America’s excellence in tech developments is already famous. It is one of the most adept countries when it comes to technology, so much so that various other countries rely on their exceptional resources. For example, China’s dependency for the past two decades on chips from the US companies has resulted in a situation where ZTE, one of the leading Chinese telecom company, is at a brink of collapse because of the US government’s trade embargo. It could be said that over-reliance on American technology crippled the functionality of this company.
Fearing similar consequences for India, Ashim Roy of CardioTrack believes that India should thoroughly use indigenous AI solutions, as that is the best way to protect the interests of the nation. In technologies involving AI, Indian companies and entrepreneurs are largely dependent on technologies from the US tech giants such as IBM, Amazon and Google. Although these frameworks provide the fastest way to launch a product in the market, Roy believes that Indian companies and government is leaving themselves and their customers at the whims of the US trade or foreign policy. In areas such as healthcare and cybersecurity, it is vitally important to develop indigenous solutions such that no foreign country can jeopardise lives and the wellness of the citizens of this country.
In his efforts towards the indigenous AI, Roy aims to start a campaign to create awareness around the same. Analytics India Magazine caught up with Roy to understand his views on the need for implementing indigenous technology, about his initiative, how he plans to implement it, and more.
Analytics India Magazine: When it comes to AI, Indian companies and entrepreneurs are largely dependent on technologies from US tech giants. Do think it leaves Indian companies and government vulnerable to vagaries of the US trade or foreign policies?
Ashim Roy: This is a very significant problem on multiple levels. Artificial intelligence, machine learning or deep learning applications are being used in many aspects of day-to-day activities and has helped in automating various process with a very high level of accuracy.
US has become a dominant player in AI because of its significant investments in research and development over the past two decades. Some of the other players are China, Israel and several European countries. Despite these developments, a large number of Indian and global AI applications are still being built on third-party AI engines and most of these AI engines are developed in the US. For instance, in India, many AI and ML applications are being developed on IBM Watson, which could be a significant problem based on the recent situation at ZTE. The Chinese telecom equipment giant, ZTE, is the latest victim of the recent trade embargo imposed by the US Department of Commerce. This embargo has stopped the shipment of components from US companies such as Qualcomm and others to ZTE and has led to a factory shutdown leaving 75,000 workers without jobs.
If US were to impose similar restrictions on IBM to license Watson, many Indian companies offering AI applications built on top of Watson will be in peril. Moreover, if these AI applications are in critical areas of healthcare and cybersecurity, lives of many Indians will be endangered.
It is, therefore, essential not to become completely dependent on partners from countries like the US or China or Europe in areas that are critical to the safety, security and well-being of the citizens of India.
AIM: How can India’s dependency on AWS/Google cloud engine raise security issues?
AR: Many problems can arise when the AI engine from a foreign partner is used. For instance, while training the AI engine, large volumes of data must be shared with the AI engine. Personal data security is on everyone’s mind since the Facebook’s misadventure with personal data, and there are many questions to be answered, such as:
- Has the AI application developer taken precautions to protect user identity?
- Has the AI application developer taken proper authorization to use and share the user data?
- Has the AI application developer taken proper measures to keep the data safe from the hands of hackers and terrorists?
- Is the AI application developer complaint to General Data Protection Regulations (GDPR), which came in to effect on 25th May 2018?
- Does the AI engine take care of all these data security issues? And others.
In general, most of the AI engines and framework companies will not provide answers to most of these questions in a way to give confidence to regulators and policymakers that personal data is being kept safe. Last two decades of playing loose with personal data at all levels – Governments, Businesses and Consumers – has led to the very serious concern about misuse of personal data and AI engines “see” a lot of data.
The Indian government has been developing its own data privacy laws. These are designed to protect consumers in case of a data breach. The regulatory framework specifies what data elements need to be protected. The AI application developer needs to ensure that the overall solution that includes the application, AI engine and all of the elements of the solution that have access to personal data adhere to these requirements. A tall order for an upstart AI entrepreneur. No wonder, one of the technology visionaries of our times, Elon Musk, thinks of AI doomsday.
AIM: How important is it to develop indigenous AI solutions when it comes to sensitive areas like healthcare?
AR: Future of India’s safety and security depends largely on its ability to develop indigenous technologies for areas such as healthcare, cybersecurity and data privacy.
Being dependent on foreign technology for mobile phones is acceptable because if the US stops Intel and Qualcomm from selling chips and stops Google from licensing Android to Micromax, consumers in India still can buy phones from Samsung or HTC. However, if IBM is banned from licensing Watson to Indian healthcare solution providers, the situation can have disastrous consequences for the well-being of the population. This can seriously impact the success of Modicare because without AI interpretation there can be no intervention.
One might find the idea that IBM, Google, Intel and Qualcomm not selling their products in India as preposterous. Think again. The only reason the US did not put a similar embargo on Huawei was because it wanted to avoid an all-out trade war with China, since Huawei is a much bigger company with deep-rooted political connections. The scenarios of a trade embargo are not far-fetched.
AIM: Indigenous AI solutions based on global research is the need of the hour. But how well-equipped are Indian companies and research institutes to achieve it? For instance, Indian IT companies may not have the infrastructure/platforms like Microsoft, AWS or Google? How can one remedy this situation?
AR: There is a huge amount of information on AI research and scholarly publications about it, in the public domain. Two of India’s IT majors – Infosys and Wipro have developed their own AI platforms. These activities are to gain knowledge, support various IT services and customer requirements. The reason many of the AI platform developers from US made the AI platform available to the global AI application development community at a low cost or no cost is the access to the data that the platform companies do not have.
Here’s how this works – Application developers using the AI platform need to constantly feed the AI engine with data to train the AI engine. The platform owner gets access to a steady supply of valuable data without having to pay for it. Which makes this a really sweet deal. Not surprisingly US IT majors are aggressively staking their positions in this wild west AI terrain.
In India, Wipro HOLMES and Infosys Nia are two notable entries when it comes to AI platforms. However, these are proprietary platforms. It would have been great if Infosys and Wipro had made them open access platforms to help develop the AI community in India.
This situation can be remedied however. All that is required is a sound AI policy framework, aggressive investment, political will and collaborative implementation. Sounds simple. Right?
AIM: What are the steps that Indian companies, government and policymakers can take to overcome this issue?
AR: Some of these steps are:
AI policy framework: Policy framework that encourages development of a complete AI platform. This will also ensure that foreign players abide by the rules to develop AI knowledge-base in India, ensure that platform infrastructure is implemented within national boundaries and in case of an adverse geopolitical situation, access to the platform cannot be denied.
Data Protection Regulations: All data sourced in India remains in India. Data is protected to ensure citizen’s privacy. Unless explicitly directed by the owner of the data, it cannot be shared with anyone else. Owner of the data cannot be denied access to their data under any circumstances.
Collaboration: Collaboration is the key to rapid progress on this front. Unless there are funds available from the government, neither academia nor the industry are likely move forward quickly and actively engage in collaboration. The current AI activities are akin to a set of silos that do not benefit from each other. This is India – Show me the money!
Investment: Government should lead the way by providing grants, debt and equity investment. Government must ensure that these funds can be easily accessed by entrepreneurs and scholars and used for the purpose it was allocated and ensure that it happens smoothly it has to eliminate red-tape and corruption. The current process of accessing government grants is archaic, inefficient and lacks transparency. Funds allocated to promote entrepreneurship in India in 2014 still has not found its way to entrepreneurs and startups.
Monitoring: Measuring progress against the policy framework and expectation of a set of outcomes is a MUST. If progress lacks, then implementation must be changed. The age-old saying that what is not measured or monitored does not work applies here as well.
Regulatory Enforcement: Without regulatory enforcement, the initiative will fail. When it comes to companies like Facebook, it is only through regulatory enforcement one can hope that the data and privacy of the Indian citizen be protected.
AIM: What are the challenges on the way when it comes to having this in place? Do we have the talent to sustain the momentum or build cloud infrastructure?
AR: One thing India does not lack is talent. However, talent alone will not solve the problem. What we lack is leadership. We also lack of strategic and policy level thinking about issues related to new technologies such as AI, ability to envision impact of AI on diverse areas from agriculture, waste management, transportation and politics. However, it is not fair to put all the blame on the government. If forums can be created where strategic thinkers from the industry can exchange their viewpoints and create awareness for policymakers (It does happen in certain industries), which leads to policy level initiatives in a reasonably short span of time (under a year), only then we can expect rapid progress. While India may not be a rich country; however, building cloud infrastructure, use of high-end CPUs and GPUs and creating a flourishing AI development environment is well within our capability. It is essential that government funding be available to support private players and entrepreneurs to offer such solutions to the AI research and development community within India. These types of easily accessible infrastructure will promote R&D in AI.
AIM: Please tell us about the campaign that you have started to create awareness about the policy issues Indian companies might face when it comes to technologies like AI?
AR: I would not say that I have started a “campaign” to create awareness about the need of indigenous AI development. However, I feel very strongly about the need for action from policymakers, industry and academia. Whenever, I get an opportunity speak about AI, I do bring up these issues about the need for policy level discussions and other emerging technologies that are shaping the world around us. Perhaps the publication of this interview will help start the “campaign”. And, I am absolutely committed to help in this process.
I do hope that policy hacks and business leaders and academia come together and join hands to develop these ideas and implement solutions that will protect the interests of the people of India and secure the country from failing trade talks and the whims of developed countries.
AIM: This campaign would also mean serious re-work and re-evaluation of government policies? How far do you think it is possible in the current context?
AR: We need to actively engage with various government agencies and convince political leaders about the urgency of the need. And, the list can go on and on. However, this will require funding to make the representation, develop the ideas further, collaborate with industry and academia, learn from global policy and technology initiatives and create an environment where everyone can see the outcome of these activities. If the initiative is for protecting the interests of the people, then it is essential that they become aware of the initiative and support it.
AIM: What is your action plan for this? Can you highlight how this will affect India’s talent and IT ecosystem?
AR: Some of the ideas about the action plan have been outlined above. I cannot say that I have the knowledge or expertise to create a complete action plan. However, I will be happy to work closely with other experts to create such a plan. There is a need for many qualified minds to come together for a cohesive AI policy. This is good news indeed for the Indian IT ecosystem because there will be a need for high caliber talent to make this happen. And most importantly, it will deliver “AI Made in India” from start to finish.
From diagnosis and monitoring of chronic diseases to robotic surgeries, here’s how artificial intelligence is reimagining the healthcare sector in India and the world
Artificial Intelligence (AI) and Machine Learning (ML) have already started making inroads into various industries. Healthcare is emerging as one of the biggest beneficiaries of the AI revolution. The technology is capable of facilitating easy and secure access to patient medical data, understanding and analysing their conditions. This ultimately helps improve accuracy and efficiency in the diagnosis and modernisation of health care practices.
An example of an elementary implementation of AI is the use of chatbots and virtual assistants that can take care of basic yet tedious tasks like registering medical records, clinical workflows and monitoring lab results – all in an automated and secure process. Another example is applying machine-learning algorithms to patient-generated data to tailor new treatment plans that will eventually help better serve individuals.
AI in healthcare: Opportunities for the world and India
According to an Accenture report published in December 2017, key clinical healthcare AI apps can create $150 billion in annual savings for the United States healthcare economy by 2026. “Growth in the AI health market is expected to reach $6.6 billion by 2021—that’s a compound annual growth rate of 40%,” says the report.
Another report by the CIS India published earlier this year, AI could help add $957 billion to the Indian economy by 2035. “…investment in AI in the Indian healthcare industry appears to be growing. For example, of the $5.5 billion raised by global digital healthcare companies In the July-September 2017 quarter, at least 16 Indian healthcare IT companies received funding,” the report said.
“State governments are also providing support to AI startups – with reports quoting the Karnataka government mobilising 2,000 crore by 2020 towards supporting the same. The Karnataka government also has a Startup Policy and Karnataka Information Technology Venture Capital Fund that can support AI startups,” it added.
A Transparency Market Research (TMR) report published in May 2017 suggests that the global healthcare automation market is growing at a CAGR of 8.8% and will touch $58.98 billion by the end of 2025, up from $28.31 billion in 2016.
Top AI implementation in healthcare
One of the biggest advantages of AI is going to be diagnosis. The technology can help industry stakeholders collate the massive health data that is available. It is estimated that more than 80% of the health data is unstructured, making it invisible to current systems, according to a PWC report.
Fortunately, technology firms like IBM and Google have already come up with solutions. Google’s DeepMind Health platform is working with clinics and health institutes across the world to implement Artificial Intelligence.
IBM’s popular AI, Watson, is using cognitive technology to process and analyse the vast data. “Watson can review and store far more medical information – every medical journal, symptom, and case study of treatment and response around the world – exponentially faster than any human. And it doesn’t just store data, it’s capable of finding meaning in it. Unlike humans, its decisions are all evidence-based and free of cognitive biases or overconfidence, enabling rapid analysis and vastly reducing – even eliminating – misdiagnosis,” according to a PWC report.
Monitoring of Chronic Conditions
Conditions like diabetes, cholesterol, fertility issues and cardiac heath are managed by regular monitoring and lifestyle changes. Chronic conditions are the single- largest burden on healthcare systems globally. Connected POC devices help generate a lot of data about the user’s body parameters. This can be combined with lifestyle information like food habits, exercise, etc, by an AI algorithm to help manage the conditions and adjust dosage of medication.
AI assisted Robotic Surgery
AI assisted robotics can guide the surgeon’s instrument during a procedure, cutting down the time required to do the surgery and reducing complications.
A lot of pathological evaluations like microscopy for infections like malaria, differential counts, etc, depend on image analysis. Similarly, finding out abnormalities in an MRI scan is done through manual analysis by a radiologist. In both the cases, AI can help by screening the image analysis to help the pathologist or the radiologist give a faster and more accurate diagnosis.
Using the fitness wearables
From Fitbit, Xiaomi Mi Band to Apple Watch, there are a number of smart fitness-focused wearables available. These fitness devices are coupled with applications that provide a deeper insight on the individual’s health on a daily basis. What AI can do is here is create an encrypted data and share it with the doctors or relevant people to help the individuals with better and personalised suggestions to help achieve their fitness goals.
The AI has the potential to help researchers create drugs as well. One of the popular names in this field is Atomwise, which uses deep learning process to reduce the time taken to discover new drugs. The six-year-old company raised more than $51 million in funding earlier this year. The company also said that it is offering over 50 molecular discovery programmes.
Even IBM is utilising its Watson AI to help accelerate drug research. “The platform allows researchers to generate new hypotheses with the help of dynamic visualizations, evidence-backed predictions and natural language processing trained in the life sciences domain. It is used by pharmaceutical companies, medical device companies and academic institutions to assist with new drug target identification and drug repurposing,” IBM explains on its website.
AI in Healthcare and India
India is also joining a growing list of the countries that are using AI in the healthcare. The adoption of AI in India is being propelled by the likes of Microsoft and a slew of health-tech startups. For instance, Manipal Hospitals, headquartered in Bengaluru, is using IBM Watson for Oncology, a cognitive-computing platform, to assist physicians discover personalised cancer care options, according to an Accenture report. For cardiac care, Columbia Asia Hospitals in Bengaluru is leveraging startup Cardiotrack’s AI solutions to predict and diagnose cardiac diseases.
“Last year the company embarked on Healthcare NExT, a Microsoft initiative which aims to accelerate healthcare innovation through AI and cloud computing. By working side-by-side with the healthcare industry’s most pioneering players, we are bringing Microsoft’s capabilities in research and product development to help healthcare providers, biotech companies and organizations across India use AI and the cloud to innovate,” said Anil Bhansali, Corporate Vice President, Cloud & Enterprise, Managing Director, Microsoft India (R&D) Private Limited.
Some of the initiatives of Microsoft India in healthcare include a Microsoft Intelligent Network for Eyecare (MINE) project where the company is working the government of Telangana for its Rashtriya Bal Swasthya Karyakram. The state government has adopted the MINE an AI platform to reduce avoidable blindness.
Microsoft also has a partnership with Apollo Hospitals to use AI for early detection of cardiac diseases. “The partnership between Microsoft and Apollo will enable to develop and deploy new machine learning models to predict patient risk for heart disease and assists doctors on treatment plans,” said Anil.
Healthi is a four-year-old Bengaluru-based digital health and wellness startup. The company uses predictive analytics, personalisation algorithms and machine learning to deliver personalised health suggestions.
“India faces a chronic disease risk burden. It is on its way to becoming the diabetic capital of the world with about 6% of the population diagnosed with the condition. A quarter of the population has high blood pressure or hypertension. Not just this, many people especially those in the age group of 25 to 40 are also being diagnosed with cardiovascular diseases (Journal of the American Medical Association (Jama) Internal Medicine). Thus, prevention and management of chronic diseases is an area where AI-led user engagement solutions can play a vital role,” said Rekuram Varadharaj, Co-founder and COO, healthi.
“India is extremely short in doctors at all levels, General Physicians to diagnose and help manage chronic conditions to specialist’s in Pathology and radiology. AI can help the doctors in faster diagnosis allowing them to focus on reviewing the data given by AI algorithms and work on complicated cases that AI cannot handle,” said Aayush Rai, Co-Founder, Inito, a Bengaluru-based start up.
As a disruptive healthcare diagnostics company, Cardiotrack has been extensively using artificial intelligence in cardiac care solutions. With an aim to significantly reduce the cost of diagnosis, the team, lead by Ashim Roy, is focused on addressing key challenges in underserved areas where an on-time diagnosis can make a life-changing difference.
Analytics India Magazine caught up with Ashim Roy, who along with Avin Agarwal and a team of 15, are driving their passion for serving the needs of people especially in Indian Tier-II cities. Having grown up in smaller cities in Uttar Pradesh and Rajasthan, Roy and Agarwal could relate to the need of having timely diagnosis, which is the key force that drives them, as well as Cardiotrack.
While Agarwal has a degree in Medical Electronics with a work experience in the Netherlands for five years, Roy is a graduate of BITS, Pilani and IIT Delhi and has worked in Australia, Canada and the US for 25 years. Roy shares, “Cardiotrack as a concept we created during numerous discussions with medical researchers at St John’s Research, the outcome of which led to a focus on improving healthcare delivery and chronic illness diagnosis in India”.
The founders connected primary care facility with AI to bring solutions focused on the needs of cardiovascular diseases, which is a prime reason of concern in many countries. Especially in India, where 60 million people suffer from cardiovascular diseases, there are less than 10,000 cardiologists. Sadly, most of the cardiologists and cardiac care facilities situated in urban centers.
Cardiotrack Platform And How Is It Revolutionising The Cardio Healthcare Industry
The Cardiotrack team is democratising healthcare by offering better diagnostics at primary healthcare centres which are more accessible to patients. Their platform has three components:
- Healthcare IoT Device, which is a patient monitor device capturing accurate 12-lead ECG data of a patient.
- mHealth App, which allows a user to see the captured ECG. Physicians can review the ECG anytime later when patient revisits, allowing to keep a track on other health vitals of the patients through a connected set of diagnostics devices such as blood pressure, cholesterol and blood glucose.
- Cloud Services And AI, that includes data storage and AI analytics.
AI At Cardiotrack
Roy explains that the AI work at Cardiotrack could be broadly categorised into two parts:
Diagnostics Aid: It pertains to their offering to cath labs. The AI interpretation is based on a neural network that compares a patient’s ECG with a database on 500,000 ECG scans that are already reviewed by cardiologists. The neural network works like a fast comparator to identify nearest match. The advantage of this approach is that AI engine finds several matches with different levels of accuracy. The accuracy of interpretation is purely a function of number of scans in the database for a given heart health condition.
Prediction: The health prediction is the next big frontier of AI. Cardiotrack has started their own research in AI prediction to bring multi-parameter analytics to help assess heart attack probability with a high level of accuracy. The model which is based on six parameters— ECG, blood pressure, blood Sugar, cholesterol, smoking and BMI, will take few months to come up with initial prototype. The basic idea is to develop a model that provides an accurate prediction of adverse heart diseases such as heart attack within few months. Roy believes that this AI approach based on real patient records is likely to yield accuracies higher than 80%, i.e. much more accurate than Framingham Score, which according to researchers from Harvard Med School is only 56% accurate.
Cardiotrack AI—Use cases
Roy is quick to share that they have more than a dozen ongoing discussions with hospitals and cath labs across in India as well as internationally. “Most promising among these are discussions with Braunwald Hospital Group, which is a new hospital offering low cost acute cardiac care through a chain of hospitals”.
Explaining how it works, Roy said, “We deploy Cardiotrack 12-lead ECG device at each of the clinics, after performing a due diligence of the primary care physicians. Our team trains the physician and medical team to ensure that information flow happens properly. When a patient complains of chest pain or other symptoms of cardiovascular disease, the primary care physician capture patient’s ECG, which is uploaded to cloud server and analysed using the AI interpretation engine. The AI interpretation is sent to the physician immediately, which typically reaches physician in less than two minutes. Based on the AI interpretation, physician takes necessary steps in intervention. In severe cases, physician forwards the patient’s ECG scan to the hospital and recommends that the patient visit the hospital immediately”.
He also added that they are working with Paras Hospital in Delhi NCR, Star Hospital in Ahmedabad and Columbia Asia in Bengaluru. “The results are remarkable in some cases. For instance in Paras Hospital where we have deployed 10 Cardiotrack devices, where in five months, we could identify 13 cases requiring Angiogram and seven cases requiring angioplasty”, he said. “We are now in discussion with Paras to expand the Cardiotrack deployment by adding 50 primary care clinics”, he added.
Roy list down some of the benefits of Cardiotrack as below;
- Benefits to Patients: It makes health diagnostics accessible to more patients.
- Benefits to Physicians: They can now better understand heart health of the patient and provide better guidance when there is a severe case.
- Benefits to Hospitals: Hospitals have been struggling to get sufficient patients to keep their cardiac catheterisation labs (CCL) busy. Most CCLs have utilisation levels of less than 25% of their operational capacity. The process of diagnosis at the primary care level and interpretation using AI improves operations at two levels
- Identifying patients and bringing severe cases to patients, just in time, increases capacity utilisation
- Cardiologists can devote more time on patient care rather than sorting out normal and abnormal patients.
Growth Story And Roadmap Ahead
“Our story is that of coincidences and lucky breaks”, says Roy. In the initial days, the two founders shared the cost of early development, and when they tried to raise funds in India, they did not receive traction from early-stage investors. Later, based on the results of field testing of early prototypes, they contacted potential investors in Singapore. Till now they have raised over $1 million in funding from multiple sources and countries.
“After initial trials, we approached Columbia Asia, Paras and Star hospitals, and Columbia Asia was a first to try out our solution”, he said.
On A Concluding Note
Roy says that while they have witnessed a phenomenal growth, a startup without challenges would not be a startup. “Challenges, constraints and competition keep us on our toes and make us innovate in every aspect of our activities. We must hire good people in the team, constantly improve upon our ideas and collaborate with academia to keep R&D cost affordable”.
He further shares that when it comes to AI, data Interpretation is just the beginning and perhaps the simplest of the tasks as long as one has large annotated datasets to train the AI engine. “The excitement of AI is in prediction and that is really close to our heart”, he said on a concluding note.
From diagnostics to therapeutics, artificial intelligence is set to change the way cardiovascular diseases are identified and treated By Marjo Johne
A diagnosis of coronary artery disease usually comes after a series of procedures, including blood tests, exercise to reproduce symptoms, an electrocardiogram, a chest X-ray and, finally, cardiac catheterization, where
a thin tube is inserted through the heart’s blood vessels to check blood flow and function in various parts of the organ.
It can take weeks, even months, for each patient to go through these gold-standard procedures. But this decades-old way of diagnosing coronary artery disease could soon be a thing of the past, thanks to artificial intelligence (AI) technology.
“With our technology, you just lay down, take off your shirt and, with a hand-held, battery-operated device, your doctor can take a three-minute recording that collects 10 million data points,” explains Don Crawford,
president and CEO of Analytics 4 Life Inc., a Toronto-based company focused on artificial intelligence-based medical devices. “This data is then sent to a cloud storage device, and from there our computers take that data and create a three-dimensional image of the heart, along with a detailed report,” explains Mr. Crawford.
“By the time the patient has their shirt back on, the physician has already received the report on their computer. And based on this report, they can let their patients know whether or not they need to go to the
The Analytics 4 Life technology, currently being tested in a dozen hospitals in the United States, is at the leading edge of artificial intelligence in cardiac care – an emerging field of science that’s set to change the way
cardiovascular diseases are diagnosed and perhaps even treated.
At the University of Nottingham in England, researchers used machine learning – where AI algorithms trained themselves – in order to find patterns to predict which patients would have their first heart attack over the next 10 years. After scanning close to 300,000 patient records, the researchers found that the AI algorithms did significantly better in predicting heart attacks than assessments based on the commonly used American College of Cardiology/American Heart Association guidelines.
In India, a start-up called Cardiotrack recently rolled out a technology platform that uses a hand-held device, cloud storage and artificial intelligence to capture and analyze electrocardiogram signals for specific
“The possibilities with artificial intelligence are truly exciting,” says Brian Golden, Sandra Rotman Chair in Health Sector Strategy at the University of Toronto (U of T) and the University Health Network (UHN) and vicedean
of professional programs at U of T’s Rotman School of Management.
“The ability to instantly recognize patterns and make sense of data from these patterns will improve diagnosis speed and quality, reduce wait times, improve health outcomes and reduce costs.”
In Canada, the stage is set for AI to transform how heart disease is diagnosed, treated and managed. And the Peter Munk Cardiac Centre (PMCC) is right in the centre of this exciting transformation, powered by digital data, ubiquitous connectivity and intelligent machines.
Armed with a $100-million donation from the Peter and Melanie Munk Charitable Foundation, the centre has forged a partnership with U of T’s Vector Institute for Artificial Intelligence, which was launched last March
to advance deep learning research and develop world-leading AI talent.
“This is the first formal health-care partnership that the Vector Institute has engaged in, and we anticipate that what we do with the Vector [Institute] will be the model for what it does subsequently with cancer, neurosciences and other health disciplines,” says Dr. Barry Rubin, Chair and Program Medical Director, Peter Munk Cardiac Centre
Together, the PMCC and the Vector Institute will build an AI team that includes a lead computer scientist, software engineers and PMCC clinicians. This team will work to identify heart problems that can be solved
through AI and machine learning.
“We will use predictive models and decision support to tailor patients’ care to their unique clinical and genomic traits,” says Dr. Rubin. “We will use natural language processing to communicate with patients in real time, no matter what language they are speaking. Taken together, this will improve the efficiency of health-care delivery, outcomes and patient satisfaction.” What will AI-supported cardiac care look like to patients and healthcare
professionals? Dr. Rubin paints a sample scenario, where doctors can remotely monitor patients with conditions such as abnormal heart rhythms and then bring to the hospital those who are identified as at risk of death or serious heart damage, based on AI algorithms that can analyze billions of biological and research data points.
“You can use the AI approach to pinpoint which patients face potentially lethal events, and bring them to the hospital before that happens,” says Dr. Rubin. “So you’re managing patients outside the hospital and using the
real value of AI to predict and prevent these lethal events.”
The great wealth and quality of digital health and research data at the PMCC are critical to the success of the centre’s AI goals, says Dr. Rubin. The PMCC recently flowed six of 47 disparate clinical and research databases
into a vast “data lake,” and it is now working to bring the remaining databases into the same central reservoir.
By integrating all this data – blood tests, clinical notes, X-rays, ultrasounds, CT and MRI scans, pathology slides and genetic information – in one location, clinicians and researchers can use AI to discover potential
causes of heart disease. These discoveries, in turn, can lead to new cures.
“For example, if we have data on 10,000 patients with a narrow heart valve, and it turns out that 2,000 of these patients have a similar gene mutation, we could determine how that gene works and develop new
therapies that would prevent the heart valve disease from ever developing in patients with that mutation,” says Dr. Rubin.
Toronto is one of the two epicentres for thought leadership in AI – the other place is Silicon Valley in California – and boasts a robust ecosystem for AI- and digital-based health-care innovations, says Ying Tam, head of health at the venture services for MaRS Discovery District, a Toronto innovation hub that connects entrepreneurs, business experts, researchers, educators and social scientists.
He points to Ontario companies such as Cloud DX and Deep Genomics, which use AI and machine learning to diagnose disease and design more targeted therapies.
While AI champions continue to innovate in health care, it will likely take years before many of these new technologies are adopted in clinical practice, says Mr. Tam. Regulations that govern medical technology are complicated, says Mr. Tam, and could work against the very nature of artificial intelligence. For instance, when an organization such as the U.S. Food and Drug Administration approves a medical solution, it does so based on a specific information package. With AI, information continues to change as the underlying algorithms learn from existing and new data
“What we do with the Vector [Institute] will be the model for what it does
subsequently with cancer, neurosciences and other health disciplines.” Dr. Barry Rubin, Chair and Program Medical Director, PMCC
Nevertheless, AI solutions have already been proven in other areas in health care. Dr. David Jaffray, Senior Scientist at Princess Margaret Cancer Centre in Toronto, points to the use of algorithms at the hospital
to automate the design of treatments for cancer patients.
“The technology is very attractive because it allows us to design the treatment sooner, treat patients sooner and even ensure that the treatment plans follow the appropriate protocols,” he says.
This level of AI-enabled efficiency can, in the future, also help hospitals to optimize the use of their resources, while ensuring the best outcomes for patients, says Dr. Rubin.
“If we had complete data on all of our patients, we could use AI to predict which patients that had heart valve surgery would stay in hospital three days or 10 days after their operation,” he says. “Using this AI-based
approach, we will be able to better plan and better utilize our health-care resources.”
Having cemented its partnership with the Vector Institute, the PMCC must now work to integrate AI into practice – an undertaking that requires a shift in mindset around patient care.
“We will need to train clinicians and students at the PMCC to work in environments where AI-based predictions will inform treatment decisions and the management of patients,” says Dr. Rubin. “There’s no question that
AI and machine learning are the future.”
Arun Jaitley and Jeff Bezos
Searching the vast Internet, I was unable to find a site or a news item where India’s Finance Minister Arun Jaitley and world’s richest man Jeff Bezos are mentioned together. And, yet as I got up this morning, both were on my mind because they have both decided to do something about making healthcare more affordable for people from low income families in two of the world’s largest democracies.
As part of the India’s 2018 Budget, Arun Jaitley announced a health insurance plan for 500 million low income citizens of India. Almost at the same time, Jeff Bezos announced low cost healthcare services to many of people in USA who are finding it increasingly difficult to get healthcare services.
Is this a coincidence? Or is this caused by super blue blood moon of 2018? We will never know.
And, look at the contrasting public health spending in these two countries – Present public healthcare spending in the USA is USD 4800 per citizen, while in India the figure is USD 26 per citizen. From public healthcare spending point of view, these two countries are almost two ends of the spectrum. It turns out that no healthcare system in any country is perfect, barring Singapore which has less population than a suburb of Bangalore. Common problems faced by healthcare planners are:
– Huge imbalance in demand (number of patients) and supply (physicians, specialists, care facility and infrastructure)
– High concentration of healthcare infrastructure in urban centers thus leaving a significant population base with very few care facilities
– Sharp increase in chronic disease patients
– High cost to the society due to avoidable invasive procedures
– Distance to diagnostics centers and tertiary care facilities
– Many avoidable deaths and huge loss of productivity due to delayed diagnosis
The root cause of these problems is the fact that significant investments have gone into technologies and infrastructure for tertiary care and decline of primary care or care close to home. Both Jaitley and Bezos have proposed plans that almost identical in their approach to address the healthcare problems in India and the USA:
– Better healthcare services at home or closer to home. In India, the emphasis is to improve primary care significantly. In the US, the approach is greater use of tele-medicine
– Success of both these plans depend on better use of technologies to deliver better care outcomes and reduce cost to the beneficiaries
What Jaitley and Bezos need is greater use of IoT and AI to make this happen. Companies like Cardiotrack and American Well are likely beneficiaries of these initiatives. These companies have demonstrated how technology can be deployed to optimally use available resources and drive down costs and provide better health outcomes.
Congratulations – @arunjaitley and @JeffBezos for your bold move to provide better healthcare to people who can least afford it. This is nation building at its best.