It is impossible to discuss the future of healthcare today without considering AI. It is a ubiquitous topic that creeps into every discussion. I had the opportunity to discuss AI with some industry leaders. But rather than discussing AI use cases, I was more curious to learn how AI can be put into production in hospitals today. The idea is to answer a simple question: What are the key requisites to put an AI program at scale in the hospital?

So, the discussion focused extensively on the “How” rather than the “Why” or the “What” of taking AI mainstream in the hospitals. Below are some of the insights that I gathered.

The most important aspect is the “Use Case” selection. Most AI programs get this wrong, at least in the context of the hospitals in India. While Use Case might be a simple word, there is a web of complexity in choosing them. Every Use Case must be selected in the organisation’s context and the current problems they are trying to solve. So, it is important to speak to the leadership in management and the department as well on the ground, like nurses, to figure out what the current problems that the hospital is facing are.

Once the Use Case has been established, it is important to understand what the measure of success would be. For example, any AI Use Case in the clinical area would require a high degree of accuracy. So, the model cannot have an accuracy of 70-80%. It needs to have an accuracy of more than 95% for it to be considered for selection. But in other administrative processes, lower accuracy should be ok. But again, it depends on the problem that you are trying to solve.

Once the Use Case is picked, it must be in a niche area or a process where the impact can be measured. It is important not to boil the ocean but look at areas where the results are measurable and the ROI can be measured. This helps in getting the buy-in within the Hospital.

Lastly, on this point, it is best to pick up Use Cases that help key departments do their job better. This will ensure that the program has backing and sponsorship,

The second key consideration is the maturity of the organisation itself. There are 4 key people involved in making decisions on new programs like AI.

• The main beneficiaries of the program- This again depends on the Use Case
• The domain experts, in the case of a Finance Use Case, the domain experts would be the key leaders responsible for a process like Account Receivables (AR)
• The technologists in the hospital could be people from technology, but also the doctors who are tech-savvy and have some experience with it
• The people who would maintain the AI infrastructure, as most of it involves high-performance computing and is usually very expensive to maintain

The last point for a successful program is ownership. The ownership of an AI Program should be with the business, not the technology team. The business, whether management or functional leadership, like medical directors or department heads. IT or Technology should always be an enabler, not the owner. Only the business can vouch for the ROI or the impact that the AI program creates, and the first few Use Cases must be successful so that the scale can be brought into the AI program for hospital-wide adoption.

Keep watching this space, in the coming posts I will discuss the key considerations for Use Case selection and give some guidelines on how to develop them.