How many of you personally know a friend/relative who had to take biopsies to test for cancer? On how many occasions have you seen such test result showing negative results? Of course, we all pray for a negative test result, but the time between the test and the result is highly stressful to say the least. Now, what if there was a reliable technology which can tell you almost immediately if taking a biopsy is needed or not. Wouldn’t it save a lot of money, time and stress?
Researchers at the University of Waterloo and the Sunny Brook Research Institute is using artificial intelligence (AI) to help detect some forms of cancer earlier and faster. The technology employs machine-learning software to analyze images of skin lesions and provide doctors with objective data. This while helps avoiding unnecessary biopsies, will also help in early detection of cancer. This is just one example of the power of Artificial Intelligence applied to Health Care.
Addressing Labor Demands and Costs
AI can address unmet clinical demands. The nature of work and employment is rapidly changing and will continue to evolve to make the best use of both humans and AI talent. For example, AI offers ways to fill in gaps amid the rising labor shortages expected in healthcare. According to an Accenture analysis, the physician shortage alone is expected to double in the next nine years. AI has the power to potentially alleviate burden on clinicians and give workers tools to do their jobs more efficiently. To get an idea of how AI will change things, let us take a look at Arterys and how it uses AI. Arterys was developed by mining a data set of more than 3,000 cardiac cases with known results, where it looked at the heart and blood flow. By being hooked to a MRI machine, Arterys can examine blood flow and MRI images to provide visual markings and assessments. It can provide an accurate picture of a heart in seconds, a process which once took an hour. That translates to one hour of time saved for doctors who can put the time to other uses like direct patient care.
Removing aspects of manual labor is just one area where AI could make things easier. Think about the countless conditions which a doctor is expected to memorize and recall their symptoms at the drop of a hat to diagnose a patient. AI is superior at memorizing massive amounts of data compared to a human, and AI can look at multiple symptoms described by a human to quickly diagnose what is wrong with a patient. And with machine learning, AI learns from its past actions, improving its accuracy over time.
In addition to assembling diagnoses and cutting out tedious labor, one of the biggest ways in which AI could save health care may be in a field which have nothing directly to do with medical care – namely, cutting down the administrative and bureaucratic costs which have sent prices spiraling. A 2015 study found that the United States healthcare system “wastes an estimated $375 billion annually in billing and insurance-related paperwork.” While politicians on both sides of the aisle have proposals on how to fix this bureaucratic problem, artificial intelligence and technology can massively slash costs. AI can process data entry, track databases, and overall handle the grunt work which today employs thousands of healthcare workers and wastes doctors’ time. And as it learns by going over the mounds of data, it will become more efficient and let the doctors take over the human side of actually caring for patients. AI doctors will not be supplanting humans anytime soon. But by removing much of the grunt work and memorization, they will make it easy for doctors to actually focus on caring for patients and saving lives. The fusion between human and machine may build a more productive and healthier society, albeit with some winners and losers.
A Bright Future for AI in Healthcare
AI is set to experience a big boom in the coming years. A recent study conducted by PWC amongst business leaders showed that about 70% of them feel that AI is going to have a major impact in the Health Sector, while Accenture’s forecast of around $6.1 Billion in AI market by 2021 suggests an 40% compounded annual growth rate. In just the next five years, the health AI market will grow more than 10x. Growth is already accelerating, as the number of healthcare-focused AI deals went from less than 20 in 2012, to nearly 70 by mid 2016. Below are examples of AI applications to health that are contributing to such growth.
1. Identifying patients and treatments
Companies like Cyft and HealthReveal analyze disparate data sources to pinpoint, and apply interventions to, the highest risk patients.
2. Streamlining administrative tasks
Qventus employs AI and Machine Learning to streamline hospital operations and improve safety in emergency rooms, operating rooms, and in-patient wards.
3. Chat Bots and Virtual Nursing
AI powered chat bots and virtual nursing assistance are one way to alleviate supply constraints, such as by widening the reach of video telehealth options. In this case, diagnosis can be powered by machine learning and then trained by artificial intelligence. One interesting example is Sensely, which provides an app that can be seen as a cross between WhatsApp and Siri that captures all the important signals about a person’s health. It has two main advantages. One, it is human-like and talks to patients naturally and the patients talk to the app as they would to a real nurse or doctor. Since most of the users are 65 or older, a population which grew up talking on the phone or in person, this appears more effective than texting. Two, although the research is inconsistent, some patients report trusting machines more than people. Because of that trust, empathy and non-judgement, many people often feel they can share more with a robot.
4. Medication Adherence
A study by the Mayo Clinic determined that 50 percent of patients have difficulty with medication adherence. Companies like AI Cure employ computer vision techniques to enable smartphones to recognize faces and medications, lowering the cost and improving the effectiveness of tracking and adherence programs. AI enabled coaching will allow a provider or coach to manage more than 1,000 patients simultaneously as opposed to 50-100.
5. Clinical data security
Medical records are very valuable to hackers and many of these records are pilfered through social engineering methods such as phishing or fraudulent phone calls. Protenus is a healthcare security company which applies AI to analyze enterprise-wide access logs and flag suspicious cases for administrator review.
Challenges Facing AI in Healthcare
If AI has so many positives, isn’t it a no-brainer to incorporate AI in healthcare as rapidly as possible? The short answer would be ‘no’ and the long answer follows. There are three main reasons why people have found it difficult to make rapid progress. One, is a Risk Aversion Mentality in Healthcare where ‘do no harm’ may cloud opportunities to ‘do more good’. Perhaps the single biggest contribution to excess cost and error in healthcare is inertia. The attitude of ‘this is how it’s always been done’ is literally killing people. Many Investors agree that the ultra-conservatism in the healthcare system, while intended to protect patients, also harms them by restricting innovation. Gavin Teo, Partner at B Capital Group and a specialist in digital health, cites “provider conservatism and unwillingness to risk new technology that does not provide immediate fee-for-service (FFS) revenue” as a major challenge for startups tackling healthcare. Teo also points out that the industry feels burned from recent experiences, such as “electronic medical records (EMR) digitization regulations, which were overhyped and resisted.”
Two, lack of curated datasets which are needed to train AI. Curated data sets that are robust and have sufficient breadth and depth for training a particular application are essential, but frequently are hard to access due to privacy concerns, record identification concerns, and regulations like HIPAA. For example, the genetic and behavioral data required for rare disease studies are not well-defined nor easily captured. And, risk factor information relating to hospital-acquired infections is frequently locked in inaccessible unstructured notes in the chart. Lastly, it takes enormous computing power to process tons of medical data.
However, the good news is that implementing AI solutions is becoming more and more relevant and possible because of the growing electronic data generation in health care and advances in cloud computing power. This has made AI research increasingly possible and has opened significant opportunities to make significant leaps in our goal towards a better Healthcare system, which CHIDS is pursuing.