Published: 30 January 2024
Two NIHR-funded researchers, Dr Patrik Bachtiger and Dr Mihir Kelshiker, write about how NIHR funding has contributed to the implementation of an artificial intelligence (AI) powered ‘smart stethoscope’. They discuss how this may benefit patients, the public and GPs, in addition to how their roles as clinical academics helped them to deploy the technology.
The benefit to patients and GPs
One in five heart failure patients are diagnosed with a heart condition after seeing their GP, while four in five are diagnosed after they are already unwell and require emergency care. Importantly, this isn’t because GPs aren’t doing their jobs; it’s because heart failure is notoriously challenging to diagnose. It has many non-specific symptoms, so is hard to spot during a short GP appointment. We identified an area where we might be able to better support patients, reduce the workload on GPs and save the NHS money. This led to our work with the smart stethoscope.
Our NIHR-funded research, the TRICORDER study, combines a traditional stethoscope with AI capabilities that extend beyond anything the human brain could pick up on. Two additional metal electrodes and a digital microphone are built into the stethoscope, which enables clinicians to measure the electrocardiogram (the electrical activity of the heart) and the phonocardiogram (the heart sounds). AI measures those wave forms and looks for specific features that are related to heart diseases, principally heart failure.
We wanted to ensure the smart stethoscope would be helpful for GPs. We listened to GPs’ feedback and involved hundreds of clinicians in our early research. For the first time we have approved clinical guidelines from NHS primary and secondary care leadership, guiding GPs what to do with the AI results. It is also supported at the level of the Integrated Care Systems in the NHS. This means that GPs have the confidence and support to use the technology, and patients at risk can be treated earlier on which improves quality of life and survival outcomes.
Working in research without a PhD
There’s a misconception that you need to have a PhD to work in research, but that’s not the case. I (Mihir) had little data science and statistical experience, but had strengths elsewhere, which have helped get me to where I am today. Research is a spectrum, where traditionally most focus and funding is on efficacy and effectiveness science, whereas we are focused on implementation science, which focuses on realising impact in clinical care at scale. This type of research is especially ideal for those from a non-traditional academic background.
If you have an ambition to work in research, we’d recommend speaking to as many people as possible. You’d be surprised how open senior researchers and clinicians are to having these conversations; they’re always wanting to find bright, hard-working, and passionate individuals to support and potentially work with. We have both had unrelenting support from Professor Nicholas Peters, cardiologist and researcher at Imperial College London (and chief investigator for TRICORDER). He has mentored and supported us in growing our careers as clinical academics – such relationships are invaluable.
Working as a clinical academic
Being a clinical academic in our field is so rewarding because you can quickly translate the impact of your academic activity into the real world. You're seeing patients one day, and the next day it's informing what you're doing as a researcher, implementing new ideas quickly and having the data come back to you, often within days. It’s a fast-paced and satisfying feedback loop. The other great thing about working as a clinical academic is the exposure and collaboration that occurs. You get to connect with very smart people in all sorts of different disciplines.
NIHR funding leading the way
Our first NIHR-funded study paved the way for this much bigger research endeavour, which involved 200 GP surgeries. This unique collaboration with GPs has really helped us recruit patients to take part in our research – we have been able to access nearly three million participants. It has been great to see our research thrive beyond the first NIHR AI award, improving access to patients who can benefit from this technology most. NIHR funding has also helped accelerate our expertise in the area of implementation science for AI technologies. In our experience, traditional research funding focuses on the earlier stages of research development, so it has been great to experience NIHR’s support for AI-driven research projects from an implementation perspective.
The potential for research in AI
There is this misconception that everyone working in AI needs to have expert data science skills, but that is not the case. While these skills are important, we need people working on implementation, ensuring that this research and technology safely and effectively benefits patients. Healthcare professionals can thrive in this space, by applying their unique understanding of clinical pathways, as well as both patients’ and clinicians’ priorities and concerns. That’s one of the great things about working with AI: the spectrum of opportunities is so broad.