Integrating AI with Telemedicine Solutions Improves Patient Care and Clinical Efficiencies

Virtual Care Improvements from AI

Clinicians are beginning to use generative AI to collect data from patients before conducting a virtual visit. In addition, AI can alert providers to changes in a patient’s health when connected to remote patient monitoring (RPM) devices.

Chatbots can perform triage and assist providers by interacting with patients prior to a virtual session.

“If you're going to be in a virtual care environment where your experience doesn't start with a doctor popping up on your computer immediately, it can start with a chatbot asking you some questions,” says Dr. Ronald M. Razmi, cofounder and managing director at Zoi Capital as well as the author of AI Doctor: The Rise of Artificial Intelligence in Healthcare. “Based on your responses, a chatbot can decide what kind of care you are going to need. If it is more intelligent, it can be trusted to provide good guidance. It can accelerate the process.”

DIVE DEEPER: Healthcare chatbots are expanding automated medical care.

Assisted Administrative Workflows with Generative AI

Generative AI can assist with coding for diagnoses in addition to drafting referrals, prior authorization letters and claim submissions. It can also generate letters to insurance companies outlining next steps for patient care, according to Razmi.

Telehealth visits require coding and documentation, and generative AI can assist healthcare providers with these tasks.

“It could look at the notes generated and automatically do the coding that needs to be done to charge for that encounter,” Razmi explains.

With AI tools documenting virtual care encounters, physicians can spend less time typing during or after a visit. If generative AI makes a mistake in coding or documentation related to a virtual care session, physicians reviewing the notes can catch it without harming patients, Razmi explains.

“That note doesn't get saved into the patient's chart until the clinician reviews the note and signs off on it,” he says.

AI eases administrative workflows prior to a virtual care session by eliminating the need for patients to fill out PDF questionnaires and intake forms, according to Elliott. It also can correct inaccurate insurance and pharmacy info in real time to reduce clinical care delays.

“Without AI, we have to rely on PDFs, uploads, e-faxes, or worse, telephonic intake, which is often a barrier to delivering seamless virtual care,” Elliott says.

Improving Remote Patient Monitoring with AI

Razmi sees the potential for digital health AI to improve RPM and enable diagnostics remotely. He cites an example in which patients can now test themselves for urinary tract infections by using a deep learning app that can analyze a scan from a strip that was dipped in a urine sample.

Meanwhile, amid a clinician shortage, AI can help providers diagnose conditions like atrial fibrillation or an abnormal heart rhythm, he says.      

“[RPM apps] use deep-learning AI and unstructured data. They have been FDA-approved, so they can be used today,” Razmi says. “They’ve been deemed to be safe and accurate enough to be used in the daily practice of medicine.”

RPM allows patients to get better treatments earlier. In addition, some services offer physical therapy for musculoskeletal diseases at home; AI can monitor a patient’s exercises and provide feedback, he says.

READ MORE: Remote patient monitoring enhances nurse workflows.

In another example, Razmi notes that an AI app can analyze an individual’s voice to detect if depression is getting worse. It can then offer instructions and advise patients to follow up with a provider.

AI tools allow clinicians to set threshold parameters for remote monitoring. AI can change a clinician’s workflow by providing critical alerts similar to those for critical lab values, Elliott explains. The tools also allow changes from baselines to occur over time; for example, if a patient’s blood pressure has been rising over weeks or months, she adds.

“It’s all about setting the right alerting thresholds and getting enough diverse data sets so that we know which thresholds matter and which do not — signal versus noise,” Elliott says.

The Potential of AI and Virtual Care

Once the FDA has approved healthcare AI tools, they can assist providers in performing patient triage and chronic disease management, Razmi says.

In the future, look for AI to handle all administrative tasks and fix incorrect data — names, addresses, insurance information, duplicate accounts and pharmacy — in real time, Elliott predicts.

Decision support applications incorporating AI will be a part of connected care. Although AI aids in clinical decision support today, in the future these tools will be part of the standard of care, Elliott says.

“AI will be used for triage and intake — think virtual AI medical assistant,” Elliott says. “This is done in a rudimentary way today, but the future could encompass many medical assistant activities that are currently inefficient and not the highest and best use of their time.”