Things & Thinks — XXXIII
The first quarter of 2022 is already over! As usual, it has been quite fast few months and a lot has happened in March when it comes to digital health. Read on for some interesting updates and tidbits!
▶️Research Digest
Reviewing digital health app governance in nine countries
This new review in Nature Digital Medicine compared various facets of digital health apps in nine countries including US, Singapore, UK and some EU countries. The results and discussions are interesting-
- The authors reviewed approaches to policy on 6 facets- Actors developing national framework; Intended use of framework; Key regulations underpinning framework; National framework for market access approval developed; Reimbursement approval framework developed and End-user interface to clinical practice/ patients.
- Most countries envision approved apps that will be available through different channels, e.g., national health portals where they exist, websites providing ‘catalogues’ or ‘directories’ of approved apps for specific domains, commercial App stores, or a combination of these sources.
- As far as quality criteria are concerned, five main dimensions emerge- Transparency, Health content, Technical content Security/Privacy and Usability.
- Current end-user interfaces are especially immature. Ideally, a clinician seeing a patient with a chronic or acute condition might have available a portfolio of favoured/vetted/approved apps that they could pick from, based on the patient’s characteristics.
- The authors conclude that while ongoing initiatives in many countries are ambitious and continue to make progress in the service of bringing better products to patients, the effectiveness of approaches in use is uncertain, as the operationalizations of criteria to date are rarely sufficiently specific to offer providers and patients the guidance they need to make evidence-based decisions about apps.
Arbitrage in Health Services Innovation
This commentary in JAMA Health Forum shares some interesting points about how private investment in health service innovation is taking advantage of slow-moving regulations and information asymmetry. Of particular note was the discussion about vertical integration, observing patient health status and prevalence of risk-adjustment algorithms.
Tracking the flows of private money into health care services can help identify both pain points and pricing errors. Documenting and disseminating that information is an essential first step toward policies turning arbitrage opportunities into efficiencies.
AI and explainability problem
This group of researchers report outcomes from their review of ‘four real-world datasets, six state-of-the-art post hoc explanation methods, and eight different predictive models, to measure the extent of disagreement between the explanations generated by various popular post hoc explanation methods’. The results are fun and quite revelatory!
Our results indicate that state-of-the-art explanation methods often disagree in terms of the explanations they output, and worse yet, there do not seem to be any principled approach that ML practitioners employ to resolve these disagreements. Furthermore, 86% of our online user study responses indicated that ML practitioners either employed arbitrary heuristics (e.g., choosing a favorite method) or just simply did not know how to resolve the disagreement problem.
▶️Digital Healthcare News
Tech in Digital Health
Teladoc partnered with Amazon to launch its telehealth service on Alexa. With the command, “Alexa, I want to talk to a doctor” on supported Echo devices, customers will be able to connect to a Teladoc call centre.
Google will embed its search and summarization tools within a health record maintained by EHR vendor Meditech. It will also roll out new healthcare tools, including a search engine feature that will show consumers available appointment times for select providers
GE Healthcare’s announced that its Edison Digital Health Platform will aggregate data from multiple sources and vendors, support clinical apps and provide a holistic view of each patient, according to the company. It also announced a partnership with Mountain View, California-based AliveCor through which electrocardiogram readings taken on the medical device company’s personal EKG device will be delivered directly into GE Healthcare’s MUSE Cardiac Management System.
Regulatory/Policy Brief
Radiology artificial intelligence company Aidoc received FDA 510(k) clearance for its tool for flagging and triaging cases of pneumothorax, or a collapsed lung, on X-rays.
Biopharma/Devices Brief
Sanofi struck a $30 million, five-year agreement with digital health company DarioHealth, a digital health firm offering similar services as Livongo and Omada.
Funding, Deals, Mergers & acquisitions
Private equity deals in the life sciences sector were worth nearly $26 billion in 2021, the highest amount in a decade. Medical devices and supplies deals were worth nearly $44 billion last year, which was also the highest value over the last decade.
UnitedHealth Group acquired LHC Group for ~$5.4 billion, combining the provider of healthcare services at home, mainly for older patients dealing with chronic illness and injuries, with UnitedHealth’s Optum unit, which manages drug benefits and offers healthcare data analytics services.
Biocon, an Indian biopharma-generic firm, agreed to purchase the biosimilars business of its fellow generic medicines maker Viatris, announcing Monday a cash-and-stock deal worth up to $3.3 billion.
Health IT vendor Allscripts is selling its hospitals and large physician practices business segment to Canada-based Harris Computer Corporation, a subsidiary of holding company Constellation Software, for up to $700 million in cash, including much of Allscripts’ legacy EHR business.
Other news
Amrita University, a private university in India, has launched a wearable health monitoring device for home use, that uses a finger clip to measure six body parameters: blood glucose, blood pressure, heart rate, blood oxygen, respiratory rate, and 6-lead ECG.
U.S. physicians who use EHRs spend an average of 1.84 hours a day completing documentation outside work hours, according to a research published in JAMA Internal Medicine.
Walmart and BetterUp team up on caregiver support tool, offering offers coaching circles with other caregivers and live virtual classes, exercises and wellness-tracking tools, as well as certified coaches and community support.
Nearly 50 million people in the U.S. had their sensitive health data breached in 2021, a threefold increase in three years, according to a POLITICO analysis of the latest HHS data.
Three out of four infusion pumps used to deliver medications and fluids to patients have cybersecurity flaws, putting them at increased risk of being compromised by hackers, according to a new study by Palo Alto Networks’ Unit 42 threat research service.
Amwell and LG Electronics are teaming up to jointly develop new digital health devices and tools, starting with hospital care in the U.S.
Eko — a maker of smart stethoscopes paired with software — announced Tuesday the launch of its newly redesigned app for cardiovascular disease screening.
📚Longread of the Month
Gary Marcus wrote a fascinating critique about deep learning and many pieces are so relevant for what we see in deep learning applications in healthcare!
Deep learning, which is fundamentally a technique for recognizing patterns, is at its best when all we need are rough-ready results, where stakes are low and perfect results optional. When the stakes are higher, though… we need to be much more cautious about adopting deep learning. When a single error can cost a life, it’s just not good enough. Deep-learning systems are particularly problematic when it comes to “outliers” that differ substantially from the things on which they are trained.
❄️Tweet of the Month
📊 Chart of the Month
Not a true chart, but works really well in understanding how to think about ‘AI is moving so fast’! via @fchollet
