From medical student to clinical consultancy: reflections on my Curistica placement
As a fourth-year medical student at Imperial College London with a passion for biomedical engineering and digital health, I've always been captivated by the potential of technology to revolutionise healthcare. This enthusiasm led me to participate in one of Imperial’s Digital Health product competitions, where my team and I presented an idea that I believed could significantly impact primary care.
Our proposal was for an AI-powered transcription tool designed to capture key details shared by patients during reception interactions, coupled with an AI triage system to help GPs organise their patient lists more effectively. We were confident in our idea's potential to streamline the often-overwhelmed primary care system. Little did I know that this competition would be the beginning of a journey that would completely transform my understanding of digital health product development, especially in the realm of AI.
Initial Assumptions vs. Reality
It was at this competition that I had the fortune of meeting Dr. Keith Grimes, the founder of Curistica, a pioneering clinical safety and digital health consultancy. Dr. Grimes listened to our pitch with interest but then provided a reality check that would prove invaluable. He highlighted the intricate challenges involved in introducing a new medical device, especially one powered by AI, and emphasised the fine line that exists between a medical device and a digital health product. This conversation was my first glimpse into the complex world of healthcare technology regulations, including concepts like the Digital Technology Assessment Criteria (DTAC) and Data Protection Impact Assessment (DPIA) that I had never considered before.
This input was invaluable in shaping how I approached the idea of digital health product development, particularly when it came to AI. I had previously thought the process was relatively straightforward: develop a good AI algorithm, build it into a product, and release it to the market. The apparent simplicity of creating a medical product fuelled my optimism about our AI tool. I imagined a smooth path from concept to implementation, with AI solving healthcare problems almost magically.
The Complex World of Digital Health Development
I wanted to learn more about this process and further my knowledge in the field, and was really pleased to be successful in applying for a placement at Curistica. Once I joined the team, my perspective on product development underwent a dramatic shift. As part of the team that interacts with clients and assists them in developing clinical safety protocols for their products, I've gained invaluable insights and real life experience in the realities of bringing an AI-powered digital health product to market.
The challenges of developing a digital health product are numerous and complex, and introducing AI into the mix adds several layers of complexity. It takes countless hours of rigorous testing, creating and maintaining detailed hazard logs, writing and continuously updating Product Requirement Documents (PRDs), and facilitating ongoing communication between engineers, data scientists, and clinical safety teams. We spend significant time identifying and removing bugs, addressing defects, and ensuring the safety and reliability of the AI algorithms at every stage of development.
Navigating the Regulatory Landscape
One of the most crucial lessons I've learned is the distinction between medical devices and digital health products, and how AI can blur these lines even further. While they may seem similar on the surface, the regulatory implications for each category are vastly different. Medical devices are subject to stringent regulations and require extensive clinical trials and certifications. Digital health products, while still regulated, may have different requirements depending on their functionality and intended use. When AI is involved, the complexity increases exponentially. Understanding where a product falls on this spectrum is critical for navigating the regulatory landscape.
I have also come to understand the licensing and regulation difficulties in this field are substantial, and AI adds another dimension to these challenges. My involvement in Curistica’s client project has allowed me to see how complex and ever-changing the regulatory environment is, with different requirements across various jurisdictions. For AI products, there are additional considerations around data privacy, algorithm transparency, and the potential for bias. Obtaining necessary approvals and maintaining compliance is an ongoing process that requires dedicated resources and expertise. It's not just about creating a great product; it's about creating a product that meets all the necessary safety and regulatory standards while also addressing the unique ethical and technical challenges posed by AI.
In medicine, products are held to a higher standard, and for good reason. Patient safety is paramount, and any tool or technology that interfaces with patient care must be rigorously tested and validated. When it comes to AI, this standard is even higher. We must ensure not only that the AI performs its intended function accurately but also that it doesn't introduce new risks or biases into the healthcare process. This high bar for quality and safety means that the development process is often longer and more involved than in other industries, especially for AI-powered solutions.
From Concept to Reality: Lessons Learned
My journey from dreaming up a new digital product to working on real-life product development healthtech projects at Curistica has completely reshaped my perspective on digital health innovation, particularly in the AI space. I now understand that bringing a new AI product to market in healthcare is not just about having a great idea or even building functional technology. It's about navigating a complex ecosystem of regulations, safety protocols, and clinical validations while also grappling with the unique challenges of AI development.
These challenges include:
1. Data quality and quantity: AI models require large amounts of high-quality, diverse data to train effectively. In healthcare, obtaining such data while maintaining patient privacy is a significant hurdle.
2. Algorithmic transparency: There's a growing demand for explainable AI in healthcare. We need to be able to understand and explain how our AI models make decisions, which can be challenging with complex algorithms.
3. Bias and fairness: AI systems can inadvertently perpetuate or even exacerbate existing biases in healthcare. Ensuring fairness across different patient populations is crucial and complex.
4. Continuous monitoring and updating: AI models can drift over time, potentially becoming less accurate or even harmful. Implementing systems for ongoing monitoring and updating is essential but challenging.
5. Integration with existing systems: Healthcare institutions often use legacy systems. Integrating AI solutions with these systems while maintaining data integrity and workflow efficiency is a significant challenge.
Looking Ahead: The Future of Digital Health Innovation
This experience has given me a deep appreciation for the expertise and diligence required in digital health development, especially when it comes to AI. While the challenges are significant, they serve a crucial purpose: ensuring that the products we create are safe, effective, and truly beneficial to patients and healthcare providers.
As I continue my studies and my work with Curistica, I carry with me a newfound respect for the intricacies of digital health innovation in the AI era. The path from idea to reality in this field is long and winding, but it's a journey worth taking. By understanding and embracing these complexities, we can create AI-powered digital health solutions that not only excite our imaginations but also meet the rigorous standards required to make a real difference in healthcare. I'm excited to continue learning and contributing to this vital area of healthcare innovation.
Sign up to Curistica’s newsletter for more thoughts on this topic. Let's connect on LinkedIn and explore the next game-changing innovations in primary care together.