Artificial Intelligence – the Holy Grail to Healthcare Innovation?
Last winter, Julia got a cough and a headache. For the third time she went to her physician, who sent her home with painkillers, thinking that it was, after all, just another case of flu. But what if Julia was different from the other flu-stricken patients that the doctor had seen previously? What if her symptoms were those of a rare cancer?
Julia might want to consult a health app for a second opinion. Such AI-driven solutions were presented at the event “DayOne Experts: Artificial Intelligence – The Holy Grail of Healthcare Innovation?” on June 7th which was organized in close partnership with the Life Science Forum Basel at the Biozentrum of the University of Basel.
Dr Martin Hirsch, co-founder of Ada Health, discussed how AI revolutionizes healthcare. Health apps powered by Artificial Intelligence (AI) are now playing Dr. House by tackling health mysteries. They ask a series of questions to patients on how they feel and produce a list of possible diagnoses. Patients can then use this information to solve their ailments themselves or show the results to their doctors to find a treatment.
In an era where medical diagnoses become increasingly personalized, how long will it take us to all have a virtual doctor in the pocket?
Ada-pting to complexity of health
‘Doctors are trained to believe that common things are common,’ explained Dr Hirsch. In traditional consultations, doctors evaluate patients’ symptoms and formulate hypotheses based on experience. Similarly, many health apps rely on machine learning technology for spotting patterns in historical data.
In contrast, Ada’s AI system ‘thinks’ about the symptoms from a set of rules and makes suggestions based on probabilities. This kind of ‘single shot learning’ does not require exposure to large sets of data and is thus valuable for dealing with new situations such as rare diseases.
Ada-mant about benefits to society
If Julia showed Ada’s diagnosis of a rare cancer to her doctor, would he follow up with expensive lab tests, or trust his intuition that it was just a flu? AI technologies can improve the relationship between doctors and patients as patients become more knowledgeable about their health conditions. Further, they would be equipped with ‘conversation starters’ during consultations with their physicians. The results of AI-powered tools should complement human intelligence, not replace it, as Martin Hirsch put it.
AI also changes the health system’s approach to data. A lot of effort currently focuses on enhancing the quantity and quality of ‘big data’ to get better health insights. However, many medical cases are single occurrences that can be missed by big data-driven machine learning. More energy should be spent on improving the rulebook of AI systems, such as adding the genetic data of patients.