Abstract
“Action without intelligence is a form of insanity, but intelligence without action is the greatest form of stupidity in the world.” —Charles Kettering
The data to decision journey in healthcare has improved tremendously in recent years—we are now interpreting vast amounts of data produced from myriad of real world settings, clinical trials, or devices which can be fed into a descriptive, diagnostic, predictive, or prescriptive engine for analysis, enabling us to make better decisions about the here and now based on what we've learned about the past but also informing us on how we can make better decisions in the future.
This is where Generative AI combined with a growing Digital Health technology ecosystem will have a huge impact on precision medicine as we train the systems on multimodal data sets, such as pathology, radiology, genomics, and EHR data. Harnessing the data from wearables to monitor and gauge the effects of therapeutic interventions, track symptoms, and give dosing recommendations and developing a more comprehensive at home testing infrastructure will in time provide a clearer picture of the state of a nation's health. This newly imagined digital health taxonomy will enhance disease stratification and patient selection which, will be invaluable to powering clinical trials for chronic and indeed, rare diseases.
However, we will fall spectacularly short if the necessary guardrails, policies, leadership, and skill sets are not embedded in diverse organizations tasked with interpreting this data. Digital health was given an opportunity to flourish during the pandemic, but we must continue with investment and adoption as it is the only viable way to empower many patients to take control of their health and for others to gain access to healthcare services. Chronic diseases contribute the most significant burdens to health systems and rare diseases bring untold hardship for patients and families looking for a diagnosis. Large language models can potentially circumvent the challenges faced with the paucity of real word data for rare diseases by generating synthetic data, images, and simulating rare disease phenotypes to expand upon existing data sets. Our Entreprecisioneur interview exemplifies LLM's use in other applications with a start up company using the technology to create a new immunotherapy to target hard-to-treat cancers. What is clear is that this technology will bring a host of new scientific talent and commercial enterprises to the market with new ideas and ways to harness the power of this technology across many applications. As ever, thinking ahead, keeping policy, ethics, privacy, and safety front and center will be the key to future success. I hope we can bring it all together and that it truly transforms patients' lives, and to echo Tom Lawry's words the future is indeed not what it used to be and that is a very good thing indeed.
