VPH Executive Committee Interview Series: Dr. Finn Catling on Clinical Engagement

The VPH Executive Committee Interview Series is designed to give visibility to the people driving progress in in silico medicine within our community.

We continue the series with Dr. Finn Catling, Academic Intensive Care Doctor, Wellcome Trust Clinical PhD Fellow at Imperial College London, and newly elected member of the VPH Executive Committee with responsibility for Clinical Engagement. In this interview, Dr. Catling shares his perspective on the barriers and opportunities for bringing computational models and AI into clinical practice, on building trust among both clinicians and patients, and on what counts as a real “success story” in healthcare innovation.

1. You were recently elected as an Executive Committee member with the role of Clinical Engagement. How do you plan to help our community bring in silico medicine technologies closer to the bedside?

The VPH members are already doing fantastic science. My main priority is to increase awareness of this work amongst clinicians and to foster collaboration. I see this as essential to ensuring our research remains relevant to patients and healthcare providers, to driving translation to the clinic and bedside, and ultimately to maximising our positive impact on society.

Specifically, we are targeting clinician groups where VPH has a particularly strong track record (such as cardiology) and proposing “presented by VPH” sessions at key clinical conferences. We also plan to appoint clinical champions across key specialties in VPH’s most-represented territories to build local clinician support networks. At upcoming VPH events, I hope that we can run “speed dating” sessions between STEM researchers and academic clinicians, promoting future collaboration and co-supervision. I also propose hosting interdisciplinary case rounds where clinicians and modellers discuss complex cases that benefit from in silico insights.

I would love to hear from VPH members who are interested in getting involved with the above, or who have ideas for other activities. Please get in touch via f.catling@imperial.ac.uk

2. Being both a clinician and researcher, what do you see as the biggest barriers and opportunities for integrating in silico medicine and AI into day-to-day clinical practice?

A huge proportion of the public (clinicians included!) now regularly use generative AI such as large language models in their lives outside of work. This has rapidly increased understanding of AI’s potential amongst healthcare professionals. I hope that we can capitalise on this enthusiasm, whilst making the distinction between AI and mechanistic modelling more clear and demonstrating the complementary role of the latter.

In terms of technical barriers, AI systems may be vulnerable to confounding and poor generalisation to new patient populations or (when trained on older datasets) contemporary patterns of care. In addition, some electronic health records do not capture rich enough data, at sufficient temporal resolution, to produce systems that make actionable treatment suggestions. However, many of the barriers are non-technical: in silico systems are much more likely to be adopted where they are designed in response to specific clinical problems and are integrated into clinical workflows. We recently wrote about these issues in the context of data-driven resuscitation.

3. Many clinicians remain unfamiliar with, or hesitant about, computational models in patient care. What advice would you give researchers to build stronger, more productive collaborations with clinical teams?

Clinicians from specialties with a strong emphasis on applied physiology may be particularly receptive to the use of mechanistic computational models. For instance, in my own specialty of Intensive Care Medicine, we often start or titrate treatment based on mental models of physiology. A computational model can then be appealing when it usefully extends a mental model, and so helps us explain complex patient responses or predict the effects of treatment. Used in this way, computational models can also have an educational role that promotes a deeper understanding of the patient. I think this is a nice counterpoint to the pessimistic narrative that AI tools diminish our ability to think critically.

4. Public awareness of AI in healthcare is growing, and patients are increasingly hearing about these technologies. Have you noticed a shift in how patients perceive and respond to the idea of AI-supported care?

I have previously discussed these issues with patients in focus groups. Those patients welcomed the potential of AI but were, of course, keen for robust testing of clinical AI systems prior to deployment, and for strong oversight afterwards. They also felt that detailed explanations of an AI system's rationale were important for the clinicians looking after patients, and that this would help them consider multiple treatment options.

However, these views may not be representative of all patients. For instance, in a recent US survey, most respondents did not trust that healthcare AI would be used in a responsible way that prevents harm. I also suspect that some patients’ opinions will differ depending on whether AI is used to directly guide care, or is just used to enhance operational efficiency. Finally, I think almost everyone finds human empathy and interaction to be of great therapeutic importance during their hospital stay, so AI systems should not replace this.

5. Building a foundation of clinical “success stories” will be a crucial step for securing future funding and expanding the adoption of in silico medicine. What is your personal “success story”?

My own research aims to improve treatment of patients with severe infections using a personalised cardiovascular model (Bayesian inversion of a 1D vascular model). By analysis of the patient’s arterial pressure waveform, this system provides information about cardiac preload, inotropy and afterload, as well as tone of the large and small arteries. There is potential for these insights to change the way that patients are diagnosed and resuscitated. I have been fortunate enough to present this work at several international conferences, including at VPH 2024 where I was honoured to receive the Best Student Award. I’m pleased to say that we have secured funding for a prospective clinical evaluation of this system, which should begin next year. I’m also supervising work which extends the system to gain additional insights from the PPG (oxygen saturation) waveform.

This interview marks another milestone of the VPH Executive Committee Interview Series, where we spotlight the insights and experiences of our members shaping the future of in silico medicine.

We look forward to sharing more perspectives from our community in the coming months and invite members to stay connected, contribute ideas, and engage in these conversations that bridge research, clinical practice, and innovation.