Artificial Intelligence & Machine Learning in Healthcare Industry:

Case Study of an Indian Healthcare Service Provider – “ Niramai “

Capstone Project

Submitted to
Indian Institute of Management, Lucknow

In partial fulfilment for the requirements of the degree

Executive Program in Business Management, 2021-22

Authors:

S No Name
1 Ms. Shweta Patel
2 Mr. Achin Kapur
3 Ms. Mansi Chitroda
4 Mr. Tushar Gawande
5 Mr. Prasanna Venkatesh
6 Ms. Rajani Bahety
7 Ms. Sakshi Agrawal
8 Mr. Sourav Mukherjee
Details
CEO & Director, Chetan Meditech Pvt Ltd-Biotek India
SAP FICO Consultant, Independent Contract
Deputy Manager at Deloitte Haskins & Sells LLP
Senior Manager- International Business Services, Medline Plc.
Head of Health & Safety, Compliance & Governance - Diageo India
Head Mid-Market & SME Relationship – Edelweiss Group
Owner & Director, CA on Web Pvt. Ltd.
Deputy Manager at Ultratech Cement

Certificate of authorship
We certify that our team as listed above is the author of this paper and that any assistance that we have received in
its preparation is fully acknowledged and disclosed in this paper. We have also cited any sources from which we have
used data, ideas, or works, either quoted directly or paraphrased. We also certify that this paper was prepared by us
specifically for this course

Submission date: 21/01/2022

Contents:

S No Description Page No
1 Introduction – What motivated us towards this project 2
2 Niramai – Non-Invasive Risk Assessment with Machine Intelligence 3
3 Methodology 5
4 Goals/Deliverables 7
5 Findings/Results – a.) Financial Management 8
b.) Economics 9
c.) Marketing 10
6 Discussion & Interview link 11
7 India specific initiatives in the field of AI and ML in Healthcare 13
8 Practical Challenges & Limitations 15
9 Recommendations & Way forward 16
10 Conclusion 18
11 Bibliography 20
“AI and ML combined with robotics and the Internet of Medical Things (IoMT) could potentially be the new nervous
system for healthcare.”
 NITI Aayog , the apex policy think tank of Government of India

What motivated us towards this project

Artificially intelligent computer systems are used extensively in medical sciences. Common applications include
diagnosing patients, end-to-end drug discovery and development, improving communication between physician and
patient, transcribing medical documents, such as prescriptions, and remotely treating patients. While computer
systems often execute tasks more efficiently than humans, more recently, state-of-the-art computer algorithms have
achieved accuracies which are at par with human experts in the field of medical sciences. Some speculate that it is
only a matter of time before humans are completely replaced in certain roles within the medical sciences. All these
motivated us to work on a project about “Niramai” which is using software, artificial intelligence and machine learning
to in a novel way to provide breast cancer screening solutions

Artificial Intelligence in healthcare is a big opportunity area in India and for rest of the world

India’s think tank NITI Aayog, put healthcare on top priority in the list of domains that need an AI push. Barriers within
the healthcare domain, particularly in rural areas of India that suffer from poor connectivity and limited supply of
healthcare personnel can get hugely benefitted with the intervention of AI. According to NITI Aayog, in India, AI
adoption for healthcare applications is expected to see an exponential increase in the next few years. The healthcare
market globally driven by AI is expected to register an explosive CAGR of 40% through 2021 and reach $6.6 billion this
year.

The think tank believes the advances in technology, and interest and activity from innovators will allow India to solve
some of its long-existing challenges in providing appropriate healthcare to a large section of its population. AI
combined with robotics and the Internet of Medical Things (IoMT) could potentially be the new nervous system for
healthcare, presenting solutions to address healthcare problems and help various governments meet mission critical
objectives. Moreover, India is the perfect petri dish for enterprises and institutions globally to develop scalable
solutions which can be easily implemented in the rest of the developing and emerging economies.

In simple terms, solve for India means to solve for 40% or more of the world.

Nothing comes without a challenge

AI-driven diagnostics comes with its own set of challenge. Firstly, data availability for different age groups, genders,
and regions is not adequate to make a smart machine learning model. The biases show up in the results, which in turn
translates to inaccurate diagnosis for that group.
While machine learning models are already getting good at helping radiologists, how legitimate are these algorithms?
Have they been thoroughly vetted by government bodies? For example, when Google AI, in partnership with the
Ministry of Public Health in Thailand, conducted deep learning experiments in a handful of clinics, they found
fundamental issues in the way the deep learning systems were deployed. Though the model improved regularly, the
challenges came from factors external to the model like Software as a medical device.

To tackle the most pressing challenges, it is recommended to focus on five principles:

1. Acknowledge that medical data can be incomplete, incorrect, missing, and biased.
2. Recognise how ML systems can result in centralising power at the expense of patients and health care
providers alike.
3. Machine learning designers must emphasise on how new systems will interface with medical systems.
4. Recognise that patients have their own expertise distinct from doctors.
5. A shift of focus from bias and fairness to focus on power and participation.

AI incorporation has to work around challenges such as shortage of qualified healthcare professionals and services
and non-uniform access to healthcare across India. India’s eHealth ambitions are yet to gain significant traction despite
promising starts with initiatives such as National eHealth Authority (NeHA), Integrated Health Information Program
(IHIP), and Electronic Health Record Standards for India.

According to NITI Aayog, in India, AI adoption for healthcare applications is expected to see an exponential increase in
the next few years. The think tank believes the advances in technology, and interest and activity from innovators will
allow India to solve some of its long-existing challenges in providing appropriate healthcare to a large section of its
population. AI combined with robotics and the Internet of Medical Things (IoMT) could potentially be the new nervous
system for healthcare, presenting solutions to address healthcare problems and help the government meet mission
critical objectives.

Non-Invasive Risk Assessment with Machine Intelligence

NIRAMAI Health Analytix is a Bangalore-based deep-tech startup addressing critical healthcare problems through
automated solutions. Came into being in 2017, Niramai developed a novel solution for detecting early-stage breast
cancer using artificial intelligence and machine learning.

Mission: To create a Universal Cancer Screening Method that can save lives.

How is traditionally breast cancer detected?

Breast cancer is the leading type of cancer in women. According to WHO, one in every 8 women is at risk of developing
a breast abnormality in her lifetime. It is well-established that early diagnosis is very critical in saving a life of a cancer
patient.
For years, breast check-ups in the country have depended on mammography to examine the human breast for
diagnosis and screening. It helps in early detection of breast cancer, typically through detection of characteristic
masses or microcalcifications. A mammogram uses a low dose of radiation to take an image of the breast. The tissue
is compressed between two plates for the best image to be taken. But a lot of women hesitate to undergo
mammographic screening because it primarily uses radiation.

Further Mammography requires high capital cost for equipment and experienced radiographers. It is recommended
once every 2 years and only to women above 45 years because it cannot identify tumors effectively for younger
women, and uses of X-rays for scanning, which can make women more susceptible to cancer if screened multiple
times. In addition, it is also very painful for the subject (about 20 pounds weight is applied on the breasts while
screening).

In sonomammography, an ultrasound uses high-frequency sound waves and converts them to an image. An ultrasound
is generally not used as the primary screening tool for breast cancer. Experts suggest that sonomammography is not
so precise. Moreover, while some radiologists are able to figure out breast abnormalities easily from
sonomammography, some are not able to do so.

Other common method of clinical exam can detect tumors only after they are large enough to be palpable.

How is NIRAMAI different and better than traditional way of breast cancer diagnostics?

NIRAMAI has developed a new cancer screening software that uses machine intelligence over thermography images
to enable a low cost, easy to use, portable solution and requires minimal human supervision. It can detect cancer at a
much earlier stage than traditional diagnostic methods and self-examination and can therefore improve survival rates.
This method of breast cancer screening is of non-contact type, painless and free of any radiation, apart from being
low-cost, and universally accessible. With this solution, women of all age groups can undergo frequent screening
without any side-effects.

The core of NIRAMAI solution is Thermalytix, a computer aided diagnostic engine that is powered by Artificial
Intelligence. The solution uses a high-resolution thermal sensing device and a cloud hosted analytics solution for
analyzing the thermal images. SaaS solution has been developed using big data analytics, artificial intelligence, and
machine learning for reliable, early and accurate breast cancer screening.

Methodology

What is Thermography?
Infrared thermography is the recording of temperature distribution of a human body by measuring the infrared
radiation emitted by the surface of that body at wavelengths between 0.8 µm and 1.0 µm. A visual heat map of the
temperature distribution on the surface of an object being imaged (for example human body) is called a thermogram.
The sensitivity in the temperature variations of up to 0.025°C can also be detected using thermal cameras or infrared
cameras, due to which even minor variations in metabolism of the body can also be detected.
Particularly for breast cancer screening, the activity in the blood vessels surrounding a developing cancer is almost
always higher than the rest of the normal breast tissue. This higher activity results in temperature changes on the
surface of the breast. Since breast tissue is part of the skin, these temperature changes and vascular changes are visible
in an infrared image of the chest region. Moreover, an abnormal thermogram is amongst the earliest signs of a pre-
cancerous or cancerous lesion of the breast.
Thermalytix:

An Advanced Artificial Intelligence Based Solution for Non-Contact Breast Screening

NIRAMAI, has developed a new technique called Thermalytix©, a novel computer-aided diagnosis solution to
automate detection of potential malignancy from thermal images. The core of NIRAMAI software consists of
novel computer algorithms based on Artificial Intelligence and machine learning that analyze thermal images
and generate quantitative interpretation. The software enables a low cost, easy to use, portable solution for
detecting breast cancer which is effective in all ages of women and does not require expert skills to operate the
system. Niramai solution uses an FDA 510(k) cleared FLIR A and E Series thermal cameras and in-house
developed Thermalytix software to generate a Thermal Analysis report and interpretation result automatically,
that a radiologist can review before making a final observation on the patient.
 The protocol for thermal image capture is as follows.
 The patient is made to relax and cool in an Air Conditioned (AC) room with approximately 22-24 deg C.
 The subject should be cooled for a minimum of 10 minutes with all clothes removed from upper body with
hands sideways on her waist and with her hair tied above
 The patient should sit straight, not slouch, take back support or touch the upper part of the body during
cooling or imaging.
 Patient should face the source of air flow during cooling such that there is direct and uniform blow of cold
air on both breasts.
 Five images of the subject are captured for (a) frontal (b) left-lateral (c) left-oblique (d) Right-lateral and (e)
Right- oblique views

Efficacy of Thermalytix as a Breast Cancer screening modality in comparison with Clinical Breast Examination
and Mammography.

This comparative study was performed on 147 patients who walked into a reputed cancer hospital in Bangalore
either for preventive screening or with a breast complaint. 93% of the patients had some symptoms for breast
cancer (pain or lump) and other subjects had come for routine preventive screening. In the protocol for this
study, prior to performing Mammography, Thermalytix and Clinical Breast Examination tests were performed
taking the consent of the patients. So, results of three tests were analyzed for every subject – Clinical Breast
Examination that found palpable lumps, Thermalytix that analyzed thermal images and Mammography. All tests
were required to give one of the three results: (a) Malignant/follow-up needed (b) Non-Malignant or Benign (c)
Inconclusive (due to heterogeneously dense breast or the test could not be done).

Below diagram gives analysis of the malignant subjects with or without palpable lumps. NIRAMAI detected all
malignant patients without lumps.

Effectiveness of Mammography and NIRAMAI as a screening test (Hospital walk-ins)

NIRAMAI Thermalytix uses the physiological and thermal changes in the tissue as a signal. Hence, NIRAMAI
screening can detect malignancy in very small tumours or even when there is no palpable lump. NIRAMAI
Thermalytix is clearly more effective than Mammography and CBE for breast screening. Unlike Mammography,
it works well even on younger women and does not have issues with breast density. Thermalytix also has lower
false positives and reduces unnecessary follow-ups of benign subjects both with lumps and without.

Diagrammatic representation of AI

The central part of the figure shows the relationship between commonly used terms in the field of AI. The arrows point
to the two categories, “Broad AI” and “Narrow AI”, where AI is applied in breast imaging. Examples of these
applications are outlined in the lists under each headingNIRAMAI Health Analytix broke its original monolithic web
services architecture into smaller pieces that could seamlessly scale in parallel. By utilizing Google Kubernetes
container orchestration system to deploy containerized applications in a managed, production-ready environment,
NIRAMAI quickly realized a new opportunity to meet its scalability challenges. The business is already running its
applications and nine ML models in containers through Google Kubernetes Engine. According to Dr.Geetha , they are
already seeing improvements with Google Kubernetes Engine because of its scalability by design. With support from
Google, NIRAMAI is aiming to make its products fully accessible and scalable within India and beyond

The business is also using the Firebase mobile application platform to run an application that enables doctors to view
and certify images on mobile devices.

Niramai Health Analytix Private Limited has its own cancer screening SaaS software that uses machine intelligence
over tomography images. The company provides hardware-software solutions, such as compact screening device with
cloud processing for specialty hospitals; handheld device with real-time cloud-based diagnostics for independent
medical practitioners; and a portable solution for large-scale screenings

Goals and Deliverables of NIRAMAI

A 2018 report of Breast Cancer statistics recorded 1,62,468 new registered cases and 87,090 reported deaths. Cancer
survival becomes more difficult in higher stages of its growth, and more than 57% of Indian women suffer from stage
3 and 4 of breast cancer at the time of detection which is highly alarming.

Risk of breast cancer increases with age, about 80% of women diagnosed with breast cancer each year are 45 or older,
and about 43% are 65 or above. Hence it is very important to detect cancer at its very early stages even before
symptoms crop up

NIRAMAI came up with the below mentioned goals and objectives

 Early detection: Detect breast cancer at its early stages through reliable patented software / machine learning
algorithms. The advantage of machine learning algorithms is it gets better and better with time
 Risk free detection: Provide radiation free, non-touch, non-painful and most importantly for women of all
ages. The current practice of detection through radiation has its own risk and aftereffects in the long run.
 Accuracy and reduction of manual intervention: Manual analysis of thermal images requires highly trained and
skilled professionals, and many times results in several false positives. However, NIRAMAI Thermalytix has
70% higher positive predictive value than Thermography
 Low-cost detection: To provide a low cost, quick, accurate, automated, portable cancer screening tool that
can be operated in any clinic. NIRAMAI is offering screening through standalone health pods at an inaugural
offer of Rs 99 and the report gets generated instantly. End to end time taken to screen a patient is
approximately 20mts and the screening time is only 5mts
  Mass outreach and scalability: Realizing the dream of running largescale cancer diagnosis test and screening
in urban, rural, and semi-urban areas
 Cancer screening at our doorsteps: Ultimately, it’s about bringing breast cancer screening to the doorsteps of
women all over the world

Having developed a proof of concept to secure market acceptance, technology and clinical validation, Niramai needed
support to take its business to the next level

Financial Management

Healthtech is the next frontier and in India, there are a million problems waiting to be solved using technology.
According to a recent KPMG report, “Even though multiple advantages are provided by healthcare start-ups, they have
not yet received a steady stream of funding to support their venture.”
Health tech start-ups tend to be service-based platforms – and these tend not to be unicorns. Other barriers that
funders face include low returns and long lead times to get those returns.
Niramai was founded in 2016 and had raised seed funding of $1 million in April 2017. Post that it has raised $6
million in Series A funding led by Japanese VC firm Dream Incubator, Beenext and other investors such as Binny
Bansal, Co-founder, Flipkart, pi Ventures, Axilor Ventures and Ankur Capital.

It is almost 5 years of operation, and the broad financials of the company is tabled below:

Rs in Crores

Items FY 2016-17 FY 2017-18 FY 2018-19 FY 2019-20

Turnover 0.00 0.16 0.61 0.75
Less: Operating Cost 0.01 1.64 3.29 8.38
EBIDTA (0.01) (1.48) (2.68) (7.63)
Other Income 0.00 0.23 1.22 2.16
PAT (0.01) (1.07) (1.36) (8.23)
Net worth 2.30 5.27 46.95 39.57
Net Fixed Asset 1.36 1.80 2.37 2.15
Investment 0.00 3.07 45.85 40.12
Inventories 0.00 0.00 0.24 0.15
Accounts Receivable 0.00 0.00 0.33 0.11
Accounts Payable 0.02 0.00 0.14 0.84
Cash Conversion Cycle (Days) - 4 262 -281

All start-ups in their initial phase, face the problem of lack of 360-degree approach and professional guidance. With
the infusion of money by Dream Incubator in the year FY2017-18, Mr. Eto Munehiko, Managing Director, Dream
Incubator joined Niramai board, bringing in the experience & professional guidance. The company is still making losses
because this kind of business has long gestation period for making returns. It is an asset light company and currently
funded with money which is lying in investment under asset side. In FY2019-20, the company is taking advantage of
using creditors money & hence the cash conversion cycle is negative.

As of now, the company has support from various investors. However, the support needs to continue given the
gestation period for getting return. Local and national governments should encourage funding by setting up start-up
hubs and a healthcare innovation fund. Meanwhile, the private sector has a dual responsibility of a guide and investor
for the development of healthcare start-ups in India.
Over and above offering funding, it is essential that the major hospital chains, pharmaceutical companies, and
diagnostic labs take charge as mentors to the healthcare start-ups.

Economics – Cost and Economies of Scale

Niramai test is more accessible and affordable than the standard breast cancer screening techniques.

The device cost is 1/10th or 1/15th the cost of a mammogram. A 2D mammography machine costs around ₹80 lakh
while a 3D machine costs above ₹2 crore. Niramai machine, which comprises several thermal sensors and the software,
costs ₹8 lakh.

For the end-user too, it is subsidized and costs one third of a mammography. The revenue model is simple: hospitals
and clinics will levy a screening charge from each patient. For bulk users like hospitals and big clinics, Niramai has a
different pricing strategy depending on their volume.

Scaling seamlessly

Niramai has over 70 installations at hospitals and diagnostic centres in India, with Apollo Clinic being their marquee
partner. Apollo Clinic has implemented Niramai’s systems at five of their 72 clinics.

With its Google Cloud Platform infrastructure scaling seamlessly, the business has grown and over the next year, plans
to expand across India. It has created two products to maximize reach – a real-time triaging report for women in rural
screening camps and a detailed report for women taking the test in diagnostic centers and hospitals in urban areas.

"With support from Google, we want to make our product fully accessible and scalable within India and beyond,"
says Dr. Manjunath.

With economies of scale, NIRAMAI’s plan is to provide testing at Rs 100 per scan as against a mammography which
costs Rs 2,500 . In case of Thermalytix

 There is no need for a doctor to administer the test as it can be done by a health worker as well
 To complete a mammogram report in hospitals and diagnostic centres signature of virologist is needed, which
adds to overall the cost of the test. Whereas it is not needed for Thermalytix.

Disruptive Innovation
Niramai Technology is not only affordable but is also simple in usage term, that large section of women can access it.

Niramai has got its technology patented for breast cancer screening. The machine learning techniques used in
Niramai’s solution have led to it winning 10 US-granted patents. It has got three patents from India, two each from
Singapore and China, and one from Canada. “The U.S. FDA has already given its nod for Niramai’s hardware, but we
are awaiting the clearance for the software,” says Manjunath.

Niramai recently received the CE mark approval, ISO 13485 and MDSAP (Medical Device Single Audit Programme)
International Certifications. The CE mark approval indicates that the product may be sold freely in any part of the
European Economic Area.
Marketing: Leaders leading from the front at NIRAMAI

The Company is working towards increasing awareness about the products among doctors as well as women who
need screening. They work closely with leading radiologists, oncologists, gynecologists, and other breast cancer
focused partners to increase awareness and confidence among their peers. They also consciously working towards
evangelization among women through communities, content, and social media. The unique experience that the
product delivers has also been a great source of positive word-of-mouth.

Dr.Geetha Manjunath Founder, Chief Technology Officer, and Chief Executive Officer, NIRAMAI Health Analytix

Decided to put the organization forward for Launchpad – A global accelerator program that helps start-ups
scale by matching them with Google's people, networks, and technologies. "We wanted to learn how to scale our
business, including our mobile and cloud infrastructure," she says.

"From a technology perspective, we wanted to accommodate traffic from the nine installations we had operating at
the time, while from a business perspective, we wanted to continue to gain acceptance from doctors and clinicians
who were cautious about adopting new technologies," she adds. "In addition, we wanted to understand what legal
and regulatory clearances and clinical trials had to be completed to expand into markets such as the United States,
Asia, and Europe."

Nidhi Mathur, Chief Operating Officer and Co-founder, NIRAMAI Health Analytix

Launchpad gave us an opportunity to be in an environment with 100 mentors from all parts of the world,"
says. "We were exposed to mentors from different industries and with highly specific skills." These mentors provided
the technical and business assistance NIRAMAI Health Analytix needed to realize its ambitions. One mentor helped
the business modify the interface to its product to be more intuitive and easier to use for the clinicians and nurses
who would be its primary users.

"With support from Google, we want to make our product fully accessible and scalable within India and beyond." Says
Dr. Geetha Manjunath.

Launchpad mentors advised NIRAMAI Health Analytix to break its original monolithic web services architecture into
smaller pieces that could seamlessly scale in parallel. "We learned about the Kubernetes container orchestration
system and how Google Kubernetes Engine could help us deploy containerized applications in a managed, production-
ready environment," says Dr. Manjunath. "We quickly realized this would help us meet our scalability challenges."

The business is running its applications and nine ML models in containers through Google Kubernetes Engine. "We've
already seen improvements with Google Kubernetes Engine because we're now growing in ways that are scalable by
design," says Dr. Manjunath. The business is also using the Firebase mobile application platform to run an application
that enables doctors to view and certify images on mobile devices.

Going beyond hospitals and Diagnostic centres

While the initial operational model was to focus on hospitals and diagnostic centers , NIRAMAI soon realized that
hospitals could be considered as simple distribution channels and there may be many more places where we could
reach our end users, who were the actual beneficiaries.

This paved the way to add new channels, so NIRAMAI established innovatively established its presence in a shopping
mall. While it's still an experiment, this initiative has enabled women who may hesitate to go to a cancer centre for
screening when they are well , to just visit the screening centre in a mall !

The equipments are easily portable. It has done several outreach camps in partnership with clinics, companies, and
non-governmental organisations.
Discussion & Interview Link with Company Heads
Questionnaire with Company Heads

In an interaction with Shweta Papriwal, Editor, IIFL, Nidhi Mathur, COO & Co-founder, NIRAMAI, said, “We provide
our solution to hospitals and diagnostic centres as well as to organizations conducting large-scale health screenings.”

1-Brief us about your business and products.

Niramai in Sanskrit means being free from illness. It is also an acronym for Non-Invasive Risk Assessment using Machine
AI. NIRAMAI has created a revolutionary technology to identify breast abnormalities in a privacy-aware, radiation-free
way that is effective for women of all age groups. Thermalytix is our patented technology that uses a high-resolution
thermal sensing device and a cloud-hosted analytics solution for analysing the thermal images using Machine Learning
and Artificial Intelligence and automatically generating detailed reports with medical parameters that can help doctors
in their clinical decisions.

NIRAMAI offers breast health screening and diagnostic solutions that are portable, age-agnostic, zero radiation, and
non-contact. These benefits very effectively address the concerns and limitations for greater adoption of screening
for women across all segments.

2-What is the process of screening breast cancer with your software? How does this software operate?

The solution attempts to remove the common inhibitions that stop women from coming forward for a breast
screening. Many Indian women feel reluctant to opt for breast cancer screening because they feel their privacy is being
invaded. But, with NIRAMAI's software, women can walk in confidently and get their breasts examined with no
touching or even seeing them. Our hardware-software technology solution, called Thermalytix, uses Artificial
Intelligence and Machine Learning to automatically analyze heat patterns of the chest to detect abnormal patterns
using remotely operated thermal sensors.

The process of screening is very simple. The thermal device is placed about three feet in front of the patient to
capture thermal readings of her chest region. The NIRAMAI software tool automatically analyses these images and
sends the analysis report for certification to radiologists.

3-What is the level of accuracy?

The solution has good accuracy of detection and has published detailed clinical studies in international peer reviewed
journals, showing clinical validity of their solution. Those studies also show effectiveness of this test in young women
for whom mammography is inconclusive due to breast density issues and highlight the ability to detect lumps so small
that they could not be felt in hand examination.

4-How did you develop this technology?

We were working on developing the technology since 2013. The team had deep competence in the field of AI and
worked on closely with radiologists and oncologists to perfect the solution and bring the accuracy level on par with
the standard of care.

5-What is your strategy of creating awareness regarding your products?

It is important for us to increase awareness about our products among doctors as well as women who need
screening. We are working very closely with leading radiologists, oncologists, gynecologists, and other breast cancer
focused partners to increase awareness and confidence among their peers. We are also consciously working towards
evangelization among women through communities, content, and social media. The unique experience our product
delivers has also been a great source of positive word-of-mouth for us.

6-Is this technology affordable and effective?

Our solution works with a wide range of thermal sensors. Today, FDA-approved thermal sensors of desired
specifications are available at a fraction of a cost of mammography machines, bringing the cost of ownership down
and enabling hospitals and diagnostic centres to offer the test at a much lower price point than mammography. With
increased volumes, the prices can be brought down even further.

7-Are you associated with hospitals/ individual practitioners?

NIRAMAI aims to expand the availability of breast health screening solutions to women, and for that purpose, we
work with various stakeholders including hospitals, diagnostic centres, individual practitioners, NGOs, Government,
and corporates. We work closely with the medical community for bringing our solution to women. Our solution is
available in more than ten hospitals and diagnostic centres in Bengaluru, Mysore, Pune, and Dehradun, and we
continue to add more locations. We provide end-to-end solutions to the institutions, including equipment, training,
report generation, and doctor certification service to those centres.

We also work closely with gynaecologists and general practitioners who are very often the first touchpoint for
women seeking medical advice regarding breast health issues.

8-Are you spreading awareness regarding breast cancer in rural areas too, where there is very less knowledge and
limited avenues?

Awareness remains low among the rural population leading to lack of emphasis on preventive screening. We
conduct and support the screening camps for the rural and urban poor in collaboration with organizations such as
Karnataka Cancer Society, outreach arms of various reputed hospitals and health-focused NGOs. In addition, we are
launching a number of initiatives to increase awareness of breast health for women and men in general.

We are also looking at setting up a mobile screening van to enable us to visit rural areas along with the equipment
and enable more women to be screened for early-stage breast cancer.

Interview Links with Company Heads

https://youtu.be/hxjwbOD2hBQ

https://youtu.be/5xFytMAThQY

India specific initiatives in the field of AI and ML in Healthcare

According to Satya Nadella, recent advances in AI have been “pretty stunning” but what the humanity is going to see
soon will be even more profound across the spectrum and health care is set to become an AI-first industry.

Microsoft CEO Satya Nadella kicked off “AI for Accessibility” — a $25 million, five-year programme for developers
globally including in India. The programme has put AI tools in the hands of developers to accelerate the development
of accessible and intelligent AI solutions for nearly one billion people with disabilities. Designed for the low-vision
community, one such Microsoft app called “Seeing AI” is harnessing the power of AI to describe people, text and
objects. It can tell visually-impaired persons what is around them.

The Manipal Group of Hospitals has tied up with IBM’s Watson for Oncology to aid doctors in the diagnosis and
treatment of 7 types of cancer. Watson for Oncology is used across its facilities, where more than 2.00,000 patients
receive cancer care each year. Here, AI is used to analyse data and research evidence and improve the quality of the
report, in turn increasing patient trust. Importantly, patients are fully aware of the process and provide their express
consent. Due care is also taken to preserve patient anonymity.

Aravind Eye Care Systems is presently working with Google Brain, after previously helping Google develop its retinal
screening system by contributing images to train its image parsing algorithms. After successful clinical trials to detect
signs of diabetes-related eye disease, it is now attempting to put it to routine use with patients.

Abbott Healthcare has used India as a testing ground for new tech innovations such as apps for the heart and liver, as
well as vertigo exercises (which use augmented and virtual reality).Pharmarack50 is a software-as-aservice (SaaS)
based application that utilises AI to automate the pharmaceutical supply chain management.

India is also host to start-up companies that specialise in harnessing AI to diagnose disease. Advenio Tecnosys detects
TB from chest x-rays and acute infections from ultrasound images. Qure.Ai uses deep learning technology to help
diagnose disease as well as recommend personalised treatment plans from healthcare imaging data, and Orbuculum
uses AI to predict diseases such as cancer, diabetes, neurological disorders, and cardiovascular diseases through
genomic data. Cureskin diagnoses six types of common skin conditions – pimples, acne, scars, dark spots,
pigmentation, and dark circles – and recommends treatment regimens through a mobile application.

According to the WHO, India is home to over five crore Indians suffering from depression and is a major contributor
to global suicides. However, seeking help for mental health issues is still stigmatised. Firms are addressing this issue
by using technology to help deal with mental health issues, usually in the form of chatbots that offer counselling while
maintaining privacy. In India, AI is being employed through chatbots such as Wysa that provide mental health support.
A person can chat anonymously with an AI-enabled system, and the chatbot is intended to provide empathetic support
and suggest practitioners to consult. Wysa is an emotionally intelligent chatbot that acts as a mental health and
behavioural coach. It uses smartphone sensors to identify and aid patients in need. Woebot is a similar service that
tracks changes in user mood on a weekly basis and finds patterns within them, offering techniques to deal with these
issues.

Medical Insurance - At present, insurers in India are limited to managing operations. Bajaj Allianz General Insurance
uses Boing, a chatbot that addresses customer queries on motor and health insurance. ICICI Lombard uses its chatbot
platform MyRA to sell insurance policies. HDFC Life’s email bot Spok claims to be first in India to automatically read,
understand, categorise, prioritise and respond to customer emails.

Google announced it has developed an AI model that can detect diabetic retinopathy with a level of accuracy on par
with human retinal specialists. Google is working on rolling out this diabetic retinopathy initiative in clinics in India
with Verily — an Alphabet-owned company which works on life sciences research and development. According to
Google, the new assistive technology can help doctors and staff screen more patients in less time, sparing people from
blindness through a more timely diagnosis.

NITI Aayog in October entered into a partnership with Microsoft to deploy AI technologies in areas such as agriculture,
health care, natural language computing and sustainable environment. In a novel effort to predict the risk of
cardiovascular disease (CVD) among the Indian population, Microsoft India and Apollo Hospitals in August launched
the first-ever AI-powered heart disease risk score API (application programme interface). Part of Microsoft’s “AI
Network for Healthcare” initiative, it will help doctors across the Apollo network of hospitals leverage the AI-powered
API to predict risk of CVD and drive preventive cardiac care across the country.The company also applied AI to devices
for early detection of diabetic retinopathy to prevent blindness. Microsoft India also announced a partnership with
SRL Diagnostics to expand the “AI Network for Healthcare” to pathology to detect cancer.

The Telangana government also adopted Microsoft Intelligent Network for Eyecare (MINE), which was developed in
partnership with Hyderabad-based LV Prasad Eye Institute. MINE uses ML and advanced analytics to predict regression
rates for eye operations, enabling doctors to pinpoint the procedures needed to prevent and treat visual impairments.
The Telangana government is already using Microsoft’s Cloud-based advanced analytics solution to screen children
from birth to 18 years of age for major conditions affecting their health.

In October, global health technology firm Philips selected 19 start-up companies for its first global start-up
collaboration programme that is focused on the application of artificial intelligence (AI) in healthcare. The programme
focuses on the application of AI-based clinical decision support tools, such as image interpretation, analysis and
integration and workflow tools, such as intelligent treatment plans for radiology, ultrasound and oncology.

IBM in November joined hands with IIT-Delhi (IIT-D) to partake in a multi-year research collaboration on AI in India,
with a focus on sectors such as healthcare and medicine

Products such as Microsoft Azure, Machine Learning, Data Analytics, CRM online and Office 365 are being used by
private healthcare providers such as Fortis Healthcare, Apollo Hospitals, L V Prasad Eye Institute (LVPEI), Narayana
Health and Max Healthcare to improve patient care.

AI and COVID-19 in India

AI and its applications have proven to be a boon in dealing with the COVID-19 situation in India. This new technology
is helpful in tracking the pandemic, screening of COVID-19 cases, containment of corona virus, contact tracing,
enforcing quarantine and social distancing, tracking of suspects, treatment and remote monitoring of COVID-19
patients, vaccine and drug development etc. A global study was conducted by PwC India and as per its report, India
witnessed the highest increase in use of artificial intelligence (AI) during COVID-19 times. As per the survey, around
73% of healthcare and pharma companies adopted AI during the year.

Practical Challenges & Limitations

Technical level

 Whilst the NHS has state-of-the-art scanners and treatments, it is also still reliant on certain record systems
that are paper-based. Thus, technological advancement is a pivotal challenge facing the NHS to allow for the
integration of new technology and the flexibility for exporting data on a mass scale.
 The NHS is also a tightly sealed system; however, companies will need access to update and modify their
algorithms. Conversely, caution is needed when opening up systems due increasing the vulnerability to
“cyber-attacks
 Despite the reported successes of AI within cancer imaging, several limitations and hurdles must be
overcome before widespread clinical adoption. With the increasing demand for CT201 and MR202 imaging,
care providers are constantly generating large amounts of data. Standards, including the Picture Archiving
and Communication System (PACS) and the Digital Imaging and Communications in Medicine (DICOM), have
ensured that these data are organized for easy access and retrieval.

Clinical level

 The curation of medical data represents a major obstacle in developing automated clinical solutions,
because it requires trained professionals, making the process expensive in both time and cost.
 The training of radiologists is also set to change with the recent incorporation of AI into the national
curriculum.
 An openness from commercial companies to disclose the limitations of their algorithms and training
radiologists how to interpret these is vital.
 Conversely, whether the adoption of such technology will require radiologists to reach a higher level of
performance to keep ahead of AI, is subject to ongoing speculation.
 Although AI can detect incidental findings that may be clinically beneficial, these findings also may be
clinically irreverent and, if not carefully framed in the correct clinical context, may increase patient stress,
health care costs, and undesired side effects from treatment.

Governance level

 The legal accountability of algorithms has been at the forefront of healthcare professionals’ questions, as no
clear guidance has been produced
 Discussions around the use of AI alongside a radiologist point towards the ultimate responsibility lying with
the clinicians, but no specifics have been detailed as to how this would fit with NHS indemnity.
 further guidance as to the accountability of the companies, reviews of “accidents” and “near misses” arising
from the use of AI should be included in department discrepancy meetings.
Recommendations and Way Forward

The government has formulated a seven-point strategy as a framework for the adoption of AI in India. This includes
developing methods for human machine interactions, ensuring safety and security of AI systems, creating a competent
workforce in line with AI and R&D needs, understanding and addressing the ethical, legal and societal implications of
AI, and measuring and evaluating AI technologies through standards and benchmarks, among others.India can learn
from the approaches of the UK and the USA in encouraging the use of AI by also focussing on:

• Formulating a Multi-Stakeholder Plan for AI in India: India has already set up an AI Task Force to come up with
recommendations for adoption of AI for economic growth. Like China, which has laid down targets for the
development of AI in phases, India must also prioritise similar clear milestones and bring in stakeholders from all
relevant sectors - including health care professionals and developers working on AI and health solutions. Working
together with industry, academia and civil society to formulate guidelines will allow the parties involved to solve the
broad and sector specific technical and ethical questions raised by AI more comprehensively and ensure that the
technologies deployed align to the national plan.

• Enabling Access to Data: By encouraging an Open Data system and ensuring that this data meets the standards set
in terms of interoperability, privacy and safety. The present use cases of healthcare AI in India tend to be very narrow
in their focus. However, AI can potentially have significant implications for macro-level, public health considerations.
For example, using AI technology to analyse the multitude of factors (including social and geographic) that could be
responsible for pre-term births can result in new insights that allow for more targeted measures to minimise the
likelihood of future occurrences. However, the biggest hindrance in implementing this is collection of relevant data in
a form that is suitable used for AI analysis. Even trial studies require a vast array of data (such as ultrasound scans,
clinical data including prescriptions, biological data such as blood tests, etc.) that must then be digitised. In addition,
the quality and uniformity of this digitised data must be ensured, which is difficult to achieve manually.

• Encouraging AI research and development: Enabling medical colleges to develop centres devoted to research on
Artificial Intelligence and facilitate the exchange of knowledge between academic centres across countries. This can
be achieved by collaborations between the government and large companies to promote accessibility and encourage
innovation through greater R&D spending. The Government of Karnataka, for instance, is collaborating with NASSCOM
to set up a Centre of Excellence for Data Science and Artificial Intelligence (CoE-DS&AI) on a public-private partnership
model to “accelerate the ecosystem in Karnataka by providing the impetus for the development of data

• Pushing for adoption of AI by businesses and the public sector: Encouraging companies to invest in AI by providing
support and incentives and encouraging the public sector to adopt AI to improve its services. A mix of models in
development financing such as IRR top up can be implemented to drive deployment. Public private partnerships are
key in addressing the lack of primary health facilities and care in rural areas. These partnerships must be carried out
under a clear framework that ensures transparency and accountability of these processes.

• Equipping existing and future labour forces with the skill sets to successfully adopt AI: Medical colleges and other
educational institutions should provide opportunities for students to skill themselves to adapt to adoption of AI in
healthcare and also push for academic programmes around AI. It is also important to introduce computing
technologies such as AI in medical schools in order to equip doctors to adopt the technical skill sets and ethics required
to use integrate AI in their practices. Similarly, IT institutes could include courses on ethics, privacy, accountability etc.
to equip engineers and developers with an understanding of the questions surrounding the technology and services
they are developing.

• Setting up a dedicated regulatory framework to oversee AI in India: At present there is no regulatory oversight in
this area and there are fears that over-regulation could lead to a stifling of innovation. This calls for a national-level
regulatory agency that oversees developments in AI in addition to formulating a framework that ensures transparency
and accountability of AI systems while promoting and enabling innovation.

• Design standards and appropriate certification system for health systems driven by AI: Proof of a clinical trial
appears to be the most common certification system asked for by hospitals and other practitioners when considering
an AI solution. Yet clinical trials are not tailored for AI technologies and are cost and time intensive. An appropriate
certification system is needed to qualify the security and quality of health systems driven by AI. Such a system can
incentivise developers to meet needed standards and can work to build trust amongst health practitioners and
patients. Design standards are also needed to encourage the development of ‘responsible AI’ Guiding principles to
encourage “responsible AI”, could include the following: 1. Transparency (operations visible to user) 2. Explainability
(process followed reaching a decision can be traced) 3. Scrutability (ability to be comprehended) 4. Credibility
(outcomes are acceptable) 5. Auditability (efficiency can be easily measured) 6. Reliability (AI systems perform as
intended) 7. Recoverability (manual control can be assumed if required)

• Regulatory and data sandbox for the health sector: Regulatory and data sandboxes have been promoted as tools
for enabling innovation while protecting privacy, security etc. Data sandboxes allow companies access to large
anonymized data sets under controlled circumstances. A regulatory sandbox is a controlled environment with relaxed
regulations that allow the product to be tested thoroughly before it is launched to the public. By providing certification
and safe spaces for testing, the government will encourage innovation in this sphere. This system has already been
adopted in Japan where there are AI specific regulatory sandboxes to drive society. Regulatory sandboxes and data
sandboxes are policy tools that can be considered for the healthcare sector to enable innovation. A governance
structure that deals specifically with healthcare can establish a system of ethical reviews of underlying health data
used to feed the AI technology along with data collected during clinical practice in order to ensure that this data is
complete, accurate and has integrity.
• Close monitoring towards harmonized implementation of EHR policy: India developed a comprehensive EHR policy
only in 2016. Comprehensive and harmonized implementation of this policy is key in enabling AI startups developing
health solutions access to accurate and usable data sets.

• Planning for the future: AI solutions in the healthcare sector are presently limited to augmenting the jobs of doctors.
Going forward, policy makers, developers, and practitioners need to consider a future where AI can make decisions
for or alongside of doctors as they shape policy, design guidelines, and best practices around use. A starting place for
this could be having ethical review committees at the organizational level that weigh the implications of the
development and use of AI solutions.

• Emphasize Privacy and Security: Health data is sensitive. AI solutions are contributing to the generation, collection,
analysis, and action of this data. AI solutions also allow for a potentially broader range of data to be collected,
combined, processed, and used in decision making. For example, a chatbot doctor will record an entire conversation
vs. a human doctor making notes during a session. The collection and storage of health related data and the potential
for bias to be reproduced through the technology raises privacy and security concerns. A key step towards ensuring
privacy and security is for India is to enact and effectively enforce a comprehensive privacy legislation.

• Continuous monitoring and research into impact of AI on the human: The impact of the use of AI on humans in
highly sensitive and personal situations has not been thoroughly studied. Understanding the impact of AI in situations
like health care where the technology can take on a partial role of caregiver will be important in developing appropriate
ethical frameworks.

Conclusion

India presently is in a unique position to be a driver in the AI and healthcare space for national and international
companies. With large amounts of data and and a burgeoning startup community, India has the opportunity to address
many health care related problems through the use of AI. In its quest for India to join the AI revolution, the government
has also undertaken a number of initiatives to drive the adoption of AI across the country.

We seem to be at the doorstep of a revolution in breast cancer screening. The developments in AI interpretation of
medical images over the last few years seem to have opened the door for incorporating stand-alone computer
interpretation of images into breast cancer screening programs. Current evidence shows that these algorithms are
approaching, if not yet have reached, expert human performance, although definitive studies that compare their
performance to actual screening results are not yet available. If and when such performance levels are achieved and
demonstrated, it seems feasible that, at least, an important reduction in the work-load for human interpretation could
be achieved, with no decrease in performance. Even if future improvements are not achieved, and therefore the
impact on performance discussed earlier remains unchanged, there might be subsequent changes down- and up-
stream that could result in an improvement in the quality of the screening program. A reduction in workload with an
unchanged AUC could allow more time for interpreting radiologists to spend on the cases that do need human review,
presumably improving accuracy. In addition, a reduction in human workload could ease the challenge of transitioning
to a more accurate but slower to interpret, imaging technology, such as DBT, again resulting in an overall improvement
in performance. Further improvements in algorithms and training sets, combined with evidence from more definite,
prospective, actual-screening-prevalence trials, could finally usher in the age of computers having a direct role in
breast cancer screening. The next few years will be very dynamic in this field.

By harnessing new insights and benefits offered by AI and Machine Learning we can increase the precision of
healthcare information to enable a new world of prediction and analysis that sees the delivery of a new citizen focus
in the care that is delivered.

Bibliography

1. https://www.accenture.com/_acnmedia/PDF-49/Accenture-Health-Artificial-Intelligence
2. https://www.thehindubusinessline.com/news/1-bn-records-compromised-in-aadhaar-breach-since-january-
gemalto/article25224758.ece
3. Central Bureau of Health Intelligence (2018), ‘National Health Profile 2018’, Ministry of Health and Family
Welfare, http://www.cbhidghs.nic.in/WriteReadData/l892s/Before%20Chapter1.pdf
4. https://www.nature.com/articles/s41416-021-01333-w#Sec15
5. https://www.indiainfoline.com/article/editorial-interviews-leader-speak/nidhi-mathur-coo-and-co-founder-
niramai-118082700167_1.html
6. https://proactive-md.com/5-ways-to-improve-the-quality-of-healthcare/
7. https://cloud.google.com/customers/niramai

8. https://www.youtube.com/watch?v=hxjwbOD2hBQ
9. http://techportfolio.net/2016/09/theres-a-30-billion-opportunity-in-indian-healthcare-tech/#ixzz7F79qa0Pf
10. https://www.mca.gov.in/mcafoportal/
11. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5813424
12. https://www.moneycontrol.com/news/business/how-niramai-is-using-artificial-intelligence-to-change-the-
face-of-breast-cancer-detection-4937181.html
13. http://www.forbesindia.com/article/isbinsight/rx-ict-digitally-disrupting-healthcare
14. ‘How Manthana uses AI to process visual medical data’, Economic Times, 17 August 2018,
https://economictimes.indiatimes.com/small-biz/startups/newsbuzz
15. ‘Maharashtra website hacked: Diagnostic lab details of 35,000 patients leaked’, The Indian Express, 3
December 2016, https://indianexpress.com/article/india/diagnostic-lab-details-of-35000-patients-leaked-
hiv-reports-4407762.
16. https://www.forbes.com/sites/insights-intela/ai-and-healthcare-a-giant-opportunity
17. https://economictimes.indiatimes.com/tech/software/how-wadhwani-brothers-sunil-and-romesh-are-using-
ai-to-serve-the-underserved
18. ‘Maharashtra CM Launches New Artificial Intelligence Centre For Rural Healthcare’, Analytics India Magazine
19. https://www.india.gov.in/spotlight/ayushman-bharat-national-health-protection-mission
20. ‘How Far Are We From Achieving Artificial General Intelligence?’, Forbes
21. ‘Artificial intelligence in health care: within touching distance’, The Lancet,
22. Ministry of Health and Family Welfare (2017), Digital Information Security in Healthcare Act
23. https://www.analyticsindiamag.com/11-indian-startupsrevolutionising-the-healthcare-sector-with-ai
24. https://cis-india.org/internet-governance/files/comments-on-draft-digital-information-security-in-
healthcare-act
25. ‘Niti Aayog plan for Aadhaar-linked digital health records raises concerns over safety and privacy’
26. NASSCOM (2018), ‘NASSCOM Artificial Intelligence Primer 2018’,
27. National Health Portal of India (2017), ‘National eHealth Authority (NeHA)’, e-Health India,
https://www.nhp.gov.in/national_eHealth_authority_neha_mtl
28. NITI Aayog (2018a), ‘National Strategy for Artificial Intelligence #AIForAll’, June 2018,
https://niti.gov.in/writereaddata/files/document_publication/NationalStrategy-for-AIDiscussion
29. ‘India: Cyber-Security: The Vulnerability Of Medical Institutions To Cyber-Attacks’, Mondaq
30. https://bigthink.com/philip-perry/how-artificial-intelligence-will-revolutionize-healthcare
31. https://isbinsight.isb.edu/digitising-indian-healthcare-records
32. “The National Digital Health Blueprint (NDHB) is just a beginning; it gives us a direction and a path to move
towards strengthening healthcare delivery systems”: Dr Harsh Vardhan’
33. https://www.moneycontrol.com/news/business/economy/healthcare-it-cos-raise-5-5-bn-in-vc-funding-in-
january-september-2017
34. ‘Cyber attack on MGM Hospital’, Asian Age
35. ‘How Microsoft Is Expanding Its Healthcare Initiative In India Using AI’, Analytics India Magazine,
https://analyticsindiamag.com/how-microsoft-is-expandingits-healthcare-initiative-in-india-using-ai
36. ‘Google Assistant is now available in eight more Indian languages’, The Times of India