From Charting to Patient Intake: The 4 Biggest Opportunities for AI in Healthcare

1. Conversational patient intake

It’s no secret: patient intake is a slog. New patients have to fill out multiple forms about their contact info, symptoms, medical history, and insurance. Sometimes, this process is fully digital and can be completed before an appointment. But it’s often also physical and in office, adding to the waiting-room stress patients often experience. And in either case, the traditional process tends to start patient-provider relationships off on a transactional note. 

With a generative AI chatbot, though, hospitals can make intake more engaging for patients. For instance, picture a patient who’s sharing their medical history with a new provider. In a chat interface, they note a history of asthma. The bot can use this information to ask if they have any relevant comorbidities (like eczema and seasonal allergies). It can also ask if they’re taking common medications for each. 

What’s more, the bot can use other information (like the patient’s age, gender, or race) to determine whether it makes sense to ask about, say, a history of heart disease or breast cancer.

Not every intake step is suited for a chatbot, of course. But the ones that are can combine to create a more conversational and personalized experience. The outcome: an intake process that engages patients, saves time, and reduces appointment-day stress.

2. Chart updates and analysis

Physicians spend an average of 18 minutes with each patient. But they spend 16 of those minutes on the patient’s electronic health record (EHR). Even when they’re actively talking with a patient about their symptoms or treatment plan, they’re constantly looking back at their screen to update the record in front of them.

Patients tend to notice this context switching. It can often feel like they aren’t getting their physician’s full attention – a problem given the already-limited face time patients have. And when patients don’t feel listened to, it’s harder to form a lasting connection with their provider.

The good news is that AI audio transcription tools can automatically capture each patient conversation in real time. And with a generative AI component, physicians can translate raw transcripts into an EHR-friendly note taking format.

After each appointment, physicians can also query a generative AI chatbot about trends across recent visits. For instance, a patient’s blood pressure might be high during their most recent appointment. But AI might note that it’s fluctuated between normal and high over the last five visits. 

At the next appointment, the physician might use this insight to ask the patient if there are any stress-inducing factors at play (like white-coat syndrome or simply a long morning commute).

Human involvement is still paramount in each AI use case: a physician needs to review each AI-generated output for accuracy. But with these tools in use, physicians can dedicate their full attention to each patient and improve the overall quality of care.

3. Dose optimization

Certain inpatient therapies require calculations in order to administer the right amount of medicine. Consider an intravenous infusion, for instance: a nurse has to calculate the drip rate and know whether to round up or down depending on the fluid involved.

With a generative AI chatbot, a nurse can find the drip rate for each infusion with a quick conversational query (e.g., “What’s the drip rate for X amount of Y solution over Z hours with a Q drop factor?”). If they need to administer several different infusions in the next hour, they can batch their calculations to save time.

Of course, it’s important to double check the output here: a rounding error could negatively impact a patient’s health. Generative AI can help with this as well – it’s often good at identifying mistakes in its own responses. The key is to build a self-check functionality or double-check prompt into the interface. This way, the AI is more likely to provide an objective review.

4. Insurance paperwork

Imagine two patients undergoing diabetes treatment at the same hospital. They both have the same insurance provider, but with different plans, which means different copays, coverage limits, and prior authorization requirements.

Hospital staff members need an easy way to understand the specifics of each insurance plan before they submit a claim. Generative AI can help. For instance, hospitals can train a chatbot on hundreds of insurance plans. Then, staff can enter a patient’s insurer and plan type and receive a comprehensive rundown in easy-to-read language. They can also ask follow-ups (e.g., “What’s the prior authorization requirement for this GLP-1 drug?”) to gain a better understanding.

With an AI tool at their fingertips, it’s easier for staff to write and submit reimbursement claims, pre-authorization forms and other required paperwork. This can increase accuracy and reduce back and forth. 

At scale, generative AI can help hospitals reduce insurance-related administrative expenses. And patients will feel the impact with fewer billing or treatment issues.