Hence the first use case would be a virtual assistant for the patients. It can give personalized and seamless experience by answering various questions without ever losing its cool or getting it wrong. This could even begin before arriving at the hospital by way of providing confirmation of the availability and timing of the doctor or department planned to visit, the way to find the exact location within the facility, getting the related background information on the doctor or department, informing of number of patients in the queue and wait times, additional information on medications, symptoms or medical conditions and so on. Further, by connecting to the information systems of the hospital, it can pull out and provide past records to the patient as and when required. One can easily imagine this saving valuable time and reducing the associated stress along with having them feel more in command and empowered over their condition and better engaged with the hospital. Additionally, this may improve adherence to treatment plans prescribed.

So how do you go about this? As I mentioned earlier, the trick employed to make ChatGPT good was to expose it to vast amount of data and train it appropriately. Hence the hospital must create a databank of as many as possible commonly asked questions with detailed answers that may be relevant and train ChatGPT. Train? How does one go about this? For this, the hospital needs to partner with companies that has expertise in the field of chatbots using natural language processing (NLP). There are many and should not be a major obstacle. Now the hospital has a customized chatbot that is a smart Jeeves and could do the job. Obviously, the question bank and related training is not a onetime activity, and it goes without saying.

The other area that comes to mind is Operational Efficiency. Lots of operations can be automated which has the potential to free up valuable resources. As an example, ChatGPT can be used in the appointment scheduling process – booking, rescheduling, cancellation and reminders. Further, with the Conversational AI in place, it can do first level of symptom check and place for review for the first human responder to have some understanding before proceeding further. The next set of automations that it can facilitate will be whenever there is a handover between departments or individuals by sending notifications that would be relevant to speedup the process. With this, the registration and admission processes would be faster, could add alerts to the nursing team as and when necessary interventions are required for the patients, facilitate smoother handover between shifts be preparing adequate notes and details to carry forward the care, reduce or eliminate data entry for repetitive tasks, manage and organize quality data, update and alert doctors on availability of patient test results as per each doctor’s preferences and so on. These are off course some of the straight forward use cases not even going into the claim management and the related processes, supply chain optimization or smart reporting to various functions, department or senior management. A natural question that would arise is why get ChatGPT to get into the picture and not leverage upon the information system in place. Among many other reasons one can think of would be reliability and scale. Additionally, once one goes down this path, there would be so much to explore further that will only increase the gap between efficiently run organizations and those otherwise. So once again the proverbial question – How do you get started? Once again, the process begins by teaming up with a good business intelligence outfit well versed with NLP and which has proven its understanding of the healthcare vertical. It certainly would not be a breeze but with one step at a time, something tangible will not be far.