Seven ways artificial intelligence can help with online doctor visits

Why it matters

Online visits (also called eVisits) let patients contact their doctor by filling out a digital form. They offer an opportunity for an online consultation with a healthcare professional, typically a GP or practice nurse, which is usually accessed through a secure online platform. An eVisit allows patients to discuss their health concerns and receive guidance or treatment recommendations without a physical appointment. Also known as e-consultations, online consultations, or online triage, they have been promoted internationally to relieve pressure on healthcare services. 

Although eVisits can increase access to healthcare for some people, they can also lead to increased staff workload, higher patient demand and potential new inequalities in access. Some of these problems might be eased with improved system design, better integration into staff workflows, and using advanced technologies like artificial intelligence (AI).   

AI tools are not yet commonly used in primary care, possibly because of barriers like data privacy and ethical concerns, legislation, and a lack of training and trust among healthcare professionals and patients.  

There is a limited amount of research available that looks at how people feel before AI is introduced, especially from the patient’s point of view, and there are still very few examples of AI being used in eVisits. These have not been studied in any depth; in our recent review, we found that 13 out of 24 eVisit systems used AI, and they mainly used it to adapt the questions patients were asked based on their previous responses. 

In our study we used qualitative methods such as interviews to try and understand what patients and staff in primary care really thought about using AI in eVisits.  

What we found 

Some patients were put off by the term ‘AI’ – they thought it was intimidating and ‘other-worldly’. The idea that AI could be used in the context of healthcare alarmed them.  

Staff on the other hand didn’t think that they knew enough about AI, and had little experience of using AI technologies in their clinical practice. Not being familiar with AI meant that patients overestimated the role it would play in their consultation, thinking that it could diagnose and prescribe without any input from their doctor or healthcare provider. They were also worried that AI might become too good and replace healthcare professionals completely. 

Overall, both staff and patients welcomed the use of AI when it was used in addition to, rather than replacing, professional clinical judgement, believing that it would be able to speed up the consultation process and reduce staff workload. At the same time, they felt that AI had the potential to both improve – and worsen – patient safety. Despite staff concerns about patients’ acceptance of AI, most patients welcomed the use of AI in eVisits.  

Patients struggled to suggest specific uses of AI themselves, while healthcare providers managed to identify seven potential uses of AI in eVisits. When these were explained to patients, they believed they were both acceptable and useful. Suggestions were: 

1. Workflow management (i.e. directing eVisits to the most appropriate professional within the practice) 

2. Signposting (for instance, to community pharmacies) 

3. Prioritising cases by highlighting urgent requests 

4. Asking relevant questions to patients 

5. Providing assistance to staff with writing letters and messages 

6. Providing self-help information 

7. Booking face-to-face appointments 

Successful implementation of AI will need clear communication from practices, demonstrating and monitoring of safety, clarifying misconceptions, and reassuring both patients and professionals that AI will not replace humans.