Say “hello!” to a combination of human and machine intelligence giving rise to predictive healthcare
Applying Data Sciences, Artificial Intelligence (AI) and human ingenuity in healthcare is the stuff legends are made of. When the COVID-19 pandemic broke in late 2019-early 2020, researchers at the University of California used Machine Learning (ML) to identify treatments. The researchers studied the entry, replication and inhibitory activity of the virus. They then used the information to sift through over 100,000 FDA registered chemicals and approved drugs and another 14 million purchasable chemicals to identify and predict which could be repurposed to treat COVID-19. Years of drug research and approvals were cut down to weeks, saving millions of lives. It does make a compelling case to support the combination of human and machine intelligence to advance medical science. Can it also be used to drive predictive healthcare that pro-actively creates interventions before people are affected by disease and need expensive care? We need look no further than the first days of the pandemic to create a water-tight case.
Kamran Khan’s Canadian AI start-up BlueDot was the first to identify an unusual flu at the market in Wuhan, China, in late 2019. This was nine days before the World Health Organization (WHO) released its official statement on COVID-19. BlueDot was set up by Khan to predict and track the outbreak of diseases and their spread. To do this, the company aggregates data on over 150 diseases around the world every 15 minutes. The data comes from sources like the Center for Disease Control, the WHO, the movement of 4 billion people on commercial flights every year, livestock health, population and climate data from satellites and from scanning over 100,000 medical articles in 65 languages published online each day. Using ML and AI, BlueDot makes projections and forecasts. It was therefore able to accurately predict eight of the first ten cities that would import the virus from Wuhan. Had health agencies been adequately alert, they could have taken pre-emptive action.
The two COVID-19 related examples sharply present the fact that the basis of medical care – for diagnostics, treatment and outcomes – is expanding rapidly. Today, it includes a range of computational frameworks, models and systems. Data mining, memetic algorithms, hybrid neuro-fuzzy models, pattern recognition, supervised, semi-supervised and unsupervised learning, the Internet of Medical Things and data-driven decision support systems are just some of the technological elements becoming common.
Most patients, when seeking physicians, look for those with “experience”. The more experienced a physician, better the chances of an accurate diagnosis, precise medication and quick recovery. Yet, even the most experienced physician must rely only on the data recorded in their experience. What if physicians, regardless of the years of practice behind them, had access to patient and treatment data accumulated by other physicians, hospitals, diagnostic clinics, insurance organizations, at the county, city, state or country level? This could open new possibilities of getting ahead of the “experience” curve. Now, physicians could use the power of collective knowledge extracted from an entire eco-system to develop treatment modalities for their patients. Analytical and automated decision-making tools could be applied to the data to come up with care regimes that go well beyond the experience of individual physicians.
The explosion of Data Gathering and mining is at the core of these changes. The pharma and health care industry generate endless data from molecular and drug research; there is vast amount of rich demographic, diagnostic and treatment data captured in electronic medical records (EHR); health care devices and monitors are spewing real-time data into cloud-based analytical engines; hospitals and care givers are generating administrative records; pharmacies have sales and prescription refill records; and medical insurance companies have claims records. Much of this data is used to meet regulatory compliances. But when brought together and massaged by Intelligent Systems (IS), it can deliver vital information to optimize and advance the cause of pre-emptive care.
The bonus of the complete “experience” creates an important inflection in the history of medical sciences. With minor alterations, Intelligent Systems could be used to monitor disease predictors and take a preventive course in preemptively treating them. This would considerably lower the pressures on the health care delivery system with improved outcomes and reduced cost of care.
There is a great need for assisting medical experts with additional data and analytical tools. A study in the US found that physicians spend just 27% of their total time with a patient. Record keeping and regulatory mandates suck up 49.2% of their time. They have little time to do additional research in their area of specialty or learn new medical processes and technologies to improve their skills. Intelligent Systems offer them a window to become more hands on in treating patients, thus boosting patient confidence.
Data sciences, automation, ML, AI and other sophisticated techniques can help doctors, care givers and hospitals become more focused in actual care delivery. For e.g. Doctors could use cohort analysis, data from patient micro-segments, known treatment options and their recorded efficacy combined with medical research to anticipate medical complications that await their patients especially in long term treatments.
While pre-emptive care is not new, it is currently limited to point-in-time solutions. These are designed to screen and/or protect patients from developing certain diseases or medical complications and help design lifestyle changes that maintain health goals, avoid medical complications and maintain the quality-of-life standards.
The debate around the cost of preventive/ pre-emptive care is well known. While some studies have shown that it may lower the cost of healthcare spending by a negligible 0.2% (US), there is the all-important patient side to the argument: If pre-emptive care buys better health than treating a disease does, is it not worth pursuing? Today, IS, AI/ML technology and Data science is shown to predict cost benefits of pre-emptive care, turning the tide against dated medical economics, spelling a new era in healthcare.
Dr Gauri Balani
Senior Director – HealthTech vertical, Marlabs LLC
Dr Gauri Balani holds a Medical degree (MBBS) and an Ivy League Master’s (MPH, YALE) and a 20 year background in the Clinical Domain. She has spent over 14 years in the Pharmaceutical vertical in senior roles in Clinical Research, Drug Safety and Analytics for various Pharma and Tech companies. She has extensive experience in companies like Lupin, Novartis, Accenture and TCS in Domain consulting. Her core strength lies in her deep understanding of technological and Data requirements in core business of Adverse Event Reporting and Regulatory Affairs.
She brings a unique combination of process, technical and operations experience, with her background in handling major outsourcing/offshoring end-to-end deals in Pharma regulatory and Healthcare space. With her adaption of Six sigma to Regulatory processes and implementing metrics for continuous improvement in the drug safety and clinical trial processes, she has a 360 perspective, including IT platforms used for these processes.