Things & Thinks XXIII
In this edition of my newsletter, I reflect upon the learnings from the COVID innovations that proved to be marginally useful at the most, followed by regulatory changes being rolled out in EU with the advent of MDR. There is the usual round-up of interesting digital health news, Tweet of the Month and Chart of the Month.
Digital health’s COVID-19 patchy track record can pave way for better solutions in future.
Since the time the pandemic began, there has been a lot of hype about digital tools and products to help, right from diagnosis through treatment and post-covid management. In each of these fields, the record though is clearly ambivalent when it comes to ultimate impact. Consider these examples-
- Contact tracing apps: As this report from NYT suggests ‘there is little rigorous research to date on whether the apps’ potential to accurately alert people of virus exposures outweighs potential drawbacks — like falsely warning unexposed people, over-testing or failing to detect users exposed to the virus.’
- ‘AI’ diagnosis of chest X-Ray: Those who read my writings know that when I see any ‘AI’ driven product, I take the ‘AI’ piece with a pinch of salt! This excellent review in Nature Machine Intelligence, authors write about ‘a nearly ideal scenario for AI to learn these spurious ‘shortcuts’’ because of the way training data is collected for these algorithms.
- Of the 26 health systems surveyed by MedCity news, nearly half used automated tools to respond to the Covid-19 pandemic, but none of them were regulated.
So what can we learn from this?
- Developers designing algorithmic decision making systems should pay more attention to the input data quality. The Nature review suggests, especially in the AI-radiology context, improved collection of training data that matches the target population and improved choice of the prediction task to involve more clinically relevant labels.
- In a brilliant commentary, Topol and Obermeyer recommend considering patient/user perspectives in the design. They write- ‘Algorithms based on datasets that pair images or waveforms with “labels” assigned by a doctor have the potential to drive improvements in efficiency and diagnostic accuracy. However, the strength of this approach can also be its weakness: by matching the performance of doctors, algorithms will also incorporate their inherent limitations.’
- Finally, there is a greater need today for regulatory agencies worldwide to start focusing on this field
Medical device regulations landscape in EU/UK is a veritable alphabet soup!
On that note, it is interesting to note two important changes in the medical device regulations that would have a direct impact on EU and UK markets but most probably also some wider implications globally.
- The European Commission’s Medical Device Regulation (MDR) entered into force in May 2017, replacing the Medical Devices Directive. The new regulation is supposed to drive safety profiles of these devices.
- An immediate fallout has been for UK and Switzerland. UK manufactured medical devices will need to comply with the MDR when placed on the EU market, the UKCA (if already applicable) has not been recognized as valid equivalent.
- In Switzerland, failure to update a Mutual Recognition Agreement ahead of the implementation of the Medical Device Regulation will mean, like post-Brexit UK, Switzerland is now a third country for the purposes of the regulation of medical devices.
- In the UK, industry groups are consulting with MHRA and hopeful that the UKCA will evolve more closely with MDR.
Unlike the MDD, the MDR imposes for the first time specific obligations on EU importers and distributors and with that makes EU an important market; so MDR will get to set a tone for global regulations governing medical devices.
Digital Healthcare news-New partners, Confusions and Funding update
- Big Tech: Amazon’s new healthcare business Amazon Care, has an enterprise client, Precor, which would use a pay-per-user model i.e. it will charge the fitness equipment manufacturer based on how many employees use its services each month. This is a departure from fixed cost contracts prevalent in this sector.
- Kaiser Permanente and Mayo Clinic are investing $100 million in a hospital-at-home company-Boston-based Medically Home.
- The FDA approved its 50th antibody in 2015, 29 years after the first one. It took just 6 more years to reach number 100, with the approval of GlaxoSmithKline’s PD1 blocker dostarlimab in April.
- Speaking of drug development, clinical trials continue to attract new digitized, siteless models. The latest to enter is CVS Health, taking it up from
- FDA gave 510(k) clearance for Bigfoot Biomedical’s insulin recommending diabetes management system, for patients with Type 1 or Type 2 diabetes who use multiple-dose injection therapy. The system’s smartpen caps take data from the FreeStyle Libre 2 continuous glucose monitor to give users insulin dose recommendations based on a physician’s instructions.
- Pharma-digital health partnerships are changing. Eli Lilly has signed compatibility agreements with four international providers including glucose monitoring sensors, digital health programs and other tools from Roche, Dexcom, Glooko and myDiabby Healthcare.
- Telehealth providers Amwell, Verizon and Doximity all launched latest virtual care offerings/platforms-Amwell’s is called Converge, Verizon’s BlueJeans Telehealth and Doximity’s Dialer.
- On 14th May 2021, Ireland’s Health Service Executive (HSE)’s computer systems, became a target for a ransomware attack.
Tidbits
- Healthcare Chart of the Month: This chart, showing Convalescent plasma’s impact (really lack of impact) in a randomized trial of 11,558 patients with COVID-19!
- Healthcare Tweet of the Month:
Well this is not really a pure healthcare tweet, but nevertheless is a great tongue-in-cheek look at what passed today as scientific literature. (Randall Munroe)