Legal and physical challenges to deep and wide adoption of AI in healthcare will have to be addressed in order to fully exploit the potential of artificial intelligence in healthcare. How to address these challenges, we discuss with Sahil Bansal, an experienced AI practitioner and co-founder of several successful startups that focus on adoption of AI in healthcare and in healthy lifestyles. This is the first part of the interview, the second part is scheduled to be published April 22, 2022
In your recent story on AI in healthcare, you have touched upon current adoptions of artificial intelligence in health: in detection and diagnostics, in health trends analysis, diet plans, chatbots, and automation. Can you elaborate on these adoptions further?
When we speak about artificial intelligence in healthcare, we are speaking about emulating the deductions that can be made after a lot of experience. In the area of diagnostics, for instance, a more experienced neurosurgeon would be able to detect common neurological issues easily and would also be able to detect rare illnesses that he has studied or heard about. A lesser experienced person would take more time to be certain of a diagnosis of a common ailment and would probably not know of potential rare illnesses. AI works on huge amounts of historic data. What we would have is the result of the cumulative experience of all these neurosurgeons, and the algorithm would be able to diagnose the patient accurately, as it would make deductions based on the data it studies. As it currently stands, today radiology is the area in which AI has made significant progress.
Similarly for health trends analysis, doctors may be able to gauge that in winters, the common cold is more common. However, since the AI algorithm would be given huge amounts of data, it might be able to detect that in winters, the people of a certain region start developing colds and spreading it to the rest of the population. With enough data, the AI would also be able to suggest if certain diet preferences in that region have led to lowered immunity levels and so on or so forth.
On the more administrative front, health care centres need to be extremely diligent and efficient. By shifting appointment scheduling to Artificial Intelligence, we could have appointments booked as per a patient’s usual time preference, or so on.
Chatbots can help patients book appointments, diagnose common illnesses, get details about billing, and so forth. This would lead to easier appointments, time saved, and simplified operations.
The scope of AI in healthcare is immense, and we’ve merely touched upon a few of the common potential use cases here.
Where do you see further areas of AI in medicine and healthcare? Where are we likely to focus in the near future?
AI makes real developments in cases where we can get meaningful and trustworthy data. So clinical trials and research, being well-documented, is an area that can benefit from the use of Artificial Intelligence.
Apart from medical research, there will be advancements in understanding genetic sequences by combining data from genomic research with literature analysis. Due to the recent pandemic, we see a lot of focus on the impact of diseases on different populations. AI could help understand how different diets and practices lead to different health dispositions among the public.
In the future, more surgeries too will be controlled by robots. However, the data required for these studies and experiments needs to be validated before it can be used. Owing to the criticality of these deductions, organisations need to ensure they handle the data correctly, which means that health care research with AI would take a considerable amount of time before it can become commonplace.
In your research, you touched upon challenges to AI adoption in healthcare: structured data, investments, and privacy. What are these challenges if you try to explain them to an outsider to the area, and how do you think we are likely to address them in the future?
AI depends entirely on data. Suppose a specific company is providing blood tests to the people of a country. If there were to be a common deficiency among the people, this would be considered highly sensitive data for the nation. We understand that the primary challenge in the adoption of AI is in maintaining privacy.
There are several regulations in place to handle this challenge, from anonymizing the data so it is not possible to track the data to specific individuals, to restricting access to this data, monitoring how the data is used, controlling where the data is saved, and how the data is encrypted.
In terms of investments, interested parties need to dedicate the immense amount of time and resources needed to make accurate results a possibility. A significant amount of time is required to ensure that the data collected is accurate and that it is understood correctly. The ramifications of AI mean that utmost diligence is expected on the part of the handlers of the data.
The second part of the interview is scheduled to be published April, 22, 2022