Sunita Tendulkar
Senior Vice President of Agile Portfolio Management & Chief of Staff, IMO Health

With more than 15 years of experience in product delivery and portfolio management, Tendulkar has been working at Intelligent Medical Objects (IMO) Health for nearly four years. She shares her advice on how healthcare organizations can leverage AI technology to get the most value from patient data.

How can AI reduce data overload for clinicians?

AI reduces clinicians’ burden by automating routine administration workflows such as clinical documentation. AI can also organize patient data into manageable summaries and surface important insights to provide clinical decision support.

What are some successful use cases of natural language processing (NLP), generative AI (genAI) and machine learning that you’re seeing in the healthcare setting?

We’re really seeing three main use cases for these technologies in the provider setting. One is intelligent messaging that can reduce the amount of time a patient or provider spends responding to communication. So, you might be in a text chat with your doctor and genAI technology is providing recommendations for your doctor on how they might respond to your question.

Another is an intelligent patient summary, which means creating a longitudinal record that pieces together different aspects of a patient’s profile to summarize what’s going on with their whole history. That saves a clinician time they might have spent combing through records to collect those pieces of information and synthesize them into that longitudinal record.

And lastly, there’s ambient AI, or technology that records patients’ and clinicians’ speech while they’re having an in-person visit and uses NLP to transform it into text. When you visit your doctor and they’re sitting behind a laptop typing the whole time, that doesn’t allow them to engage with you as a person. Ambient AI allows that clinician to spend more time with the patient and reduces the burden they would otherwise face of turning all that speech into a meaningful summary of the encounter.

How can a vendor partner help foster trust in AI as healthcare providers embark on AI-driven initiatives for their organizations?

Well, how does any industry or vendor foster trust? A lot of that just comes from transparency. For AI technology, that might mean transparency about what type of AI you’re using, what data was used to train your model and how you’ve confirmed that the data wasn’t biased. Can you explain the logic and the reasoning that’s being used for your models, the heuristics and analysis that the AI is doing, and defend and adjust it based on real-world results?

And secondly, specifically within healthcare, security and privacy are huge considerations, as well as HIPAA compliance. Patient records contain sensitive information, even outside of compliance issues, information that patients might not want shared for personal reasons. That’s an element that’s really under scrutiny when thinking of AI technology. So, as a vendor, being able to speak to the protocols you’re putting in place so that your AI is not only responsible, but also secure, is very important.

What advice would you give to healthcare organizations that are just now looking into adopting AI?

Firstly, “AI” in general is a broad term and the technology is evolving very rapidly. So, as a healthcare organization getting into that space, I think it’s important to know exactly what problem you’re trying to solve. Then you’ve got to find the specific implementation of AI that’s appropriate for you from a cost, workflow, implementation and scalability standpoint. Those are all traditional criteria that would be evaluated when making any kind of technology choice. However, specifically with AI, because it is still emerging, it’s important to find the right fit.

Secondly, I think the other aspect to keep in mind is that, again, maybe unique to healthcare, you’re dealing with decisions that could literally have life-or-death impact on a patient. And so, while certainly we want to be aggressive and innovative at times, it’s also important to take that risk into consideration. And in some cases, maybe we want to lean on more mature technology, things that might have less impact since the technology is still evolving, compared to using it on the clinical-decision workflow since that can be very impactful to a patient’s outcome. So, you really want to make sure that as an organization, you’re engaging with your clinicians and your vendors to decide what kind of technology is most appropriate for each case.