Primary care The technology

AI in primary care – The Healthcare Professional perspective

Anastacia Simonchik

<span id="hs_cos_wrapper_name" class="hs_cos_wrapper hs_cos_wrapper_meta_field hs_cos_wrapper_type_text" style="" data-hs-cos-general-type="meta_field" data-hs-cos-type="text" >AI in primary care – The Healthcare Professional perspective</span>

New automated channels for patients taking the first contact with healthcare can add value and solve major workflow issues for healthcare and patients alike. It is easy however to champion technology when one is familiar with it. What do healthcare professionals really think of automating patient contact with advanced technology? Are the benefits evident and would such tools be trusted? And, when it finally lands on their plate, what are their expectations? To explore these questions, Visiba Care conducted interviews with healthcare professionals, to get a feeling of what healthcare currently think of automated first patient contact.

Which challenges can an automated first patient contact solution solve for healthcare professionals? The biggest challenge in every healthcare professional’s mind, including our interviewees, was time. Medical professionals don’t have nearly as much time as they wished, and a large chunk of it is consumed by administrative and repetitive tasks. The situation becomes even worse for the first line of healthcare – primary care centres. Sherif Sabry, GP at a primary care centre in Sweden noted that 30% of visits in Swedish primary care concern skin conditions, which in turn are routinely signposted.

Meanwhile, the NHS is already short of 100,000 staff which is projected to reach 250,000 in the next 20 years [1]. At the same time, as mentioned in our previous article, demand for easier and quicker access is growing. This situation creates a straining combination for the first line of healthcare, as resources shrink and demand increases. Administration stays the same (or even grows) and routine tasks, such as history taking and signposting, increase. Nurses that work with phone consultations, spend half or more of their time giving administrative guidance to patients.

Another issue was noted from one of our interviewees, GP Olivia Landén.

We see a lot of patients that a good pharmacist or a good nurse could have helped much better than we can. (…) That’s one of the issues we’ve seen, that people demand to talk to a doctor about a pain where we know that it’s much better for them to talk to a physiotherapist.

As Olivia mentioned, patients are more eager to take their care in their own hands and make their own decisions, even though they may not be the right ones medically. Even though the patients’ proactivity can be frustrating in that instance, we should leverage this underexplored power with the right tools to turn it into a positive outcome in the long run.

What do healthcare professionals expect from automated first patient contact?

Our interviewees were keen to express what they would like such tools to do. Their most prominent expectation was rather grounded: They want a solution that can effectively signpost patients to the right care path – and do it with accuracy. Nevertheless, as the discussion unfolded, the standard was climbing higher. The interviewed healthcare professionals added more building blocks. Olivia Landén expands on the matter to describe the need for tools supporting healthcare professionals down the care chain:

What I would like the tool to do is to be a triage tool where you can see all the red flags “this is urgent, this is serious, this needs urgent care or this could wait for a few days – you can follow simple advice and see where it goes. It directs you to the right resource and is able to red flag– pick out something really serious.

Mats Halldin, GP at Sophiahemmet in Stockholm and medical manager at netdoktor.se, an interactive online medical knowledge base for patients, adds another layer:

This kind of system can reduce the risk for loss of information and it’s independent of the doctor, their competence or how they’re feeling that day. From the patient’s perspective, I think it’s also beneficial to be able to go through the symptom and history questions without time pressure.

On another note, Sherif Sabry shed light to another possible benefit. As he explains, while taking the symptoms, such a tool can be a great reminder of the health impact of lifestyle choices, such as smoking or alcohol consumption, and it can provide patients with valuable self-education.

When you ask someone about alcohol, you remind them of how much they drink, you remind them that it’s related to symptoms. It will be a great benefit for prevention and maybe it will sort symptoms alongside lifestyle choices. It gives silent advice!

Regarding ample functionality for an automated triage solution, the consensus is that at full development, healthcare professionals deem that such tools can be an effective signposting tool, with the ability to assess the severity of each situation and provide personalised advice.

What are healthcare professionals’ concerns regarding the use of AI?

Artificial Intelligence is a concept that can raise concerns by definition. However, a good number of them comes from either lack of information regarding the technology itself, or from a list of unknown parameters that arise mostly in Black Box technologies.

Have you explored the different AI approaches in healthcare? Read a comprehensive breakdown here.

Our interviewees noted 5 critical bottlenecks – points that would hinder their trust and usage of an AI solution.

1) Transparency of the result

Healthcare professionals noted that, much like it’s expected in their everyday practice, they want to be able to retrace signposting decisions and possible diagnoses. They want to see how the solution reached to that result and provide feedback on the accuracy of the given diagnosis or signposting. As we’ve unpacked in a previous article, not all core technology choices enable this, that is why an understanding of what is at the core of the solution is crucial.

2) Complement not substitute to healthcare professionals

In alignment with their expectations, healthcare professionals wish for effective signposting rather than immaculate diagnoses – even though indicative diagnosis are welcomed by some, making it a matter of preference. The ability of automated first patient contact to diagnose the given conditions should work as a vessel towards safe and effective triage and advice, rather than an end in itself.

3) Ask only relevant questions

When asked about building trust in patients, our interviewees expressed a strong conviction: automated solutions should gather the patient’s history in simple, accessible language and avoid irrelevant questions. Olivia Landén remarks:

If I type that my throat hurts and it asks me if I have problems peeing, that’s completely irrelevant. If it was a doctor, you wouldn’t trust that doctor, you’d think this person doesn’t know what they’re doing. (…) That’s what I think can make patients lose trust – if it’s asking irrelevant things.

4) Prevent excessive focus on efficiency

Mats Halldin attempted to echo the general feeling among healthcare professionals. According to his assessment, there is a risk that such solutions can make healthcare staff feel that too much pressure is placed on them to perform efficiently. The concern is that, with automated contact, the patient-practitioner relationship may become distant or make the personnel obsolete. This point should be taken into account with product development. Mats reflects:

It’s not about creating a distance, having more patients or creating efficiency. It’s a way of handling the demands of patients in a good way.

5) Strong ethical considerations and clear vision

On a higher level, another perspective highlighted in our discussion was the question of why. Thoughts like how is this tool developed and why?, Is the focus on the patient?, What’s the purpose of creating such a solution? should be worked through, addressed and answered with transparency, honesty and mere facts by developers and suppliers.

Will healthcare professionals trust automated first patient contact and AI?

Finally, we asked our healthcare professionals whether they would trust an AI solution in patient contact. The response was unanimous: Yes – under certain conditions. To their praise and regardless of their own positive or negative stance, healthcare professionals seem to place the standard for automated tools really high.

First of all, they require all the critical points above to be addressed. They’re anticipating a reliable tool that will perform the task it’s assigned at the utmost level of performance. The will to provide patients with easier access to healthcare is there: They want to encourage patients to be proactive, and this can be done only with the right tools.

Secondly, they want to be able to understand how the tools work. Healthcare professionals want to be able to see the human side behind the technology; They need to trust that the presented tool works on a transparent, retraceable level and is based on expert medical knowledge – not random pattern recognition.

Finally, they want the patients to be able to trust the solution as well. Healthcare professionals want the patients to have a seamless experience, without unnecessary time spent on automated communication, if it is not something that can help them. They also want the patients to be able to receive concrete and trustworthy help at the pace that patients require.

Our reflections on what we hear from the healthcare professionals are that, while that may seem like a tall order, it is the standard in technology, in innovation and specialised software development. Most importantly, the requirements and expectations above are finally something that technology and experience design have a strong capability to deliver

Quotes have been edited for clarity and length.

Visiba Group AB
Adolf Edelsvärds Gata 11 Göteborg, 414 51
Phone: 0761993666