Things & Thinks XXVI
I read a couple of interesting pieces with different perspectives about the changing nature of healthcare research and I summarize them in this edition of my monthly newsletter. These summaries are followed by the usual round-up of digital health news, and the ‘Monthly Pick Series’ with Chart, Tweet and Longread of the Month.
Happy Reading!
Healthcare Research during the Pandemic I-Public Health Crisis and Clinical Trials
I have written in the past about the nature of evidence gathering in healthcare research and how during public health crisis such as the current pandemic, legacy systems expose the gaps in our processes, methods and analyses.
This month, Scott Gottlieb-former commissioner at USFDA-wrote presciently about these gaps and what can be done to research methodology in future. Although the article is written from a US viewopoint, most of its findings and recommendations are equally applicable around the world. Here’s a summary of his thoughts-
- Poor Methodology: He cites that more than 90% of the clinical trials underway that evaluated different medicines for the treatment of COVID-19 were designed in a fashion that would never enable the studies to yield firm results that could reliably shape the practice of medicine.
- Lack of Coordination: In a crisis, resources for conducting research are scarce. Clinicians do not have time to run multiple, rigorous studies, and enrolling patients in a setting where clinicians are delivering emergency care can be difficult. When COVID-19 struck, the uncoordinated system for clinical research meant that many of the most important clinical questions did not get addressed in a timely manner.
- Focusing on the problems at hand: The US spent too much time conducting multiple studies to investigate questions that should have been more firmly answered earlier during the pandemic.
Gottlieb does not stop with pointing these problems, he also gives a broad prescription for future of clinical trials-
- In the absence of conclusive information, findings from speculative studies, observational data, and anecdotes were able to fill the ensuing information vacuum.
- The US should empower the FDA and the National Institutes of Health (NIH) to steer patients into trials that are more likely to yield actionable evidence that can inform clinical practice and improve outcomes.
- The FDA and the NIH should have the authority to prioritize among therapies those that have the most promise and accelerate enrolment of studies to evaluate these high-priority medicines.
Healthcare Research during the Pandemic II- AI Tools to catch COVID!
This is another area that I have closely observed and reviewed. The one hype that caught on everyone’s fancy during the pandemic was about the use of machine learning (or as some put it ‘AI/ML’) to detect COVID aimed at assisting the healthcare system to detect the disease earlier, faster and in a more accurate manner.
This article by Will Douglas Heaven, published on MIT Technology Review, sums up the end result-almost all such tools failed! Here are some fascinating insights-
- Laure Wynants, an epidemiologist at Maastricht University in the Netherlands, studied 232 such predictive tools and found that not a single one of them was fit for clinical use. In a similar analysis by another researcher Derek Driggs, out 0f 415 published tools promising to predict patient risks from medical images, none were fit for clinical use.
- The culprits behind this failure are not new-poor quality data (mislabelled, unknown/non-existent sources or both), ‘Frankenstein datasets’ (which are spliced together from multiple sources and can contain duplicate), incorporation bias, and lack of transparency about the codes and algorithms.
- The way out is also not totally new; Heaven acknowledges that real-world data especially the ones collected during a public health emergency can be messy. Researches such as Driggs suggest that the simplest move in this case would be for AI teams to collaborate more with clinicians. Instead of developing new models, researchers can also focus more on adopting and improving on existing models.
- Bilal Mateen, another researcher quoted, has an interesting way to put this
Until we buy into the idea that we need to sort out the unsexy problems before the sexy ones, we’re doomed to repeat the same mistakes.
Digital Healthcare news-
- Tech in Digital Health: Google famously closed its Google Health division, prompting many in the industry react about the difficulties tech giants face in healthcare. On the other hand, it also announced new partnership with SignalPath with Verily, to build its clinical trial platform ‘Baseline’, for decentralized and hybrid trials . Verily also announced partnership with Mayo Clinic, focus on a set of cardiovascular and cardiometabolic conditions. Mayo Clinic will provide clinical information, while Verily will apply its analytics and product development services to create insights that can be integrated into the provider workflow.
- Policy Brief: UK’s National Institute for Health and Care Excellence (NICE) announced that it will simplify its approach to enable earlier and equitable patient access and increased use of broader evidence base, incl real-world data. US FDA announced approval of tacrolimus to prevent lung transplant rejection, based on non-interventional (observational) study, proving the growing influence of real-world data for regulatory use cases. China announced sweeping reforms to its personal data protection regulations; on the other hand, Indian regulations will go through more consultations.
- Medtronic India entered into a strategic partnership with Statis Health, a unit of Stasis Labs, to promote the latter’s bedside patient monitoring system in India.
- Zoom’s mobile browser client would allow patients to simply click a link sent from their providers to join a telehealth appointment, nixing the need to download an app.
- Fitbit announced launch of Charge 5, with a standout feature- Fitbit Premium’s new Daily Readiness Score that takes into account users’ health metrics to guide them towards a workout geared to their body’s needs or to a day of rest.
📊Healthcare Chart of the Month:
This chart can be explained, reviewed and thought about through so many perspectives!
🔗Healthcare Tweet of the Month:
Actually two tweets this time!
📘Healthcare Longread of the Month:
Not exactly a new read, but definitely worth reading and re-reading-Ed Young has done some fantastic writing during the pandemic but his article ‘When the Next Plague Hits’ written in Aug 2018 is healthcare/science writing at its best.