Tag Archives: #deepmind

Fortnightly Healthtech Update #4

I love hearing about startups in emerging nations – I think there is so much potential for cross pollination. Here are five from India focused on blood supply, preeclampsia, kidney health, ECG, and chronic disease detection, among other applications.

The NHS in England shows good results with a pilot for the early detection of sepsis, plans a nationwide roll-out.

A fortnight ago I wrote about Deepmind’s progress in detecting kidney failure. Here’s a bit of pushback, citing some challenges for AI in healthcare in general. This phrase jumped out at me: “(clinicians) also rely heavily on human judgment to diagnose…a level of subjectivity that would be nearly impossible to program into every AI algorithm”. It was well argued almost 10 years ago by very respected clinicians that removing some subjectivity and standardizing care processes (free registration required) would help deliver higher quality care at lower cost.  In other industries, we grow productivity by selectively substituting technology for labor. And we do that in healthcare too. In pockets. Just not enough yet to make high quality, lower cost care repeatable at scale. More on that a little further down the page….

Deepmind, on the other hand, is a half a billion dollar deep hole for Alphabet by the way…

Talking of pushback, Visibly has had to pull its mobile app eye test from the market. The American Optometric Association has apparently lobbied hard against Visibly. Taking a big step back, this might be something of a moral dilemma that we need to figure out over the next 10-20 years. The cost of healthcare is still rising. If that continues, an increasing number of people in the US are going to be priced out of conventional healthcare (i.e consulting a doctor, one on one). Telehealth could be an option.  But pushing the argument further, what happens when people can’t afford to see a doctor at all, even virtually. Are we really going to deny them the opportunity to have their health evaluated solely by an algorithm? Even if that offers a less comprehensive examination than a doctor. If the only options you can afford are diagnosis by algorithm, or no diagnosis at all, which would you choose…?

The American Heart Association is launching a pilot with LifePod Solutions to help care for people with chronic cardiac conditions at home. Also in the remote patient monitoring game, Biotricity reports strong quarter to quarter growth.

Rumors abound that all is not happy in the world of Apple health. But, that hasn’t stopped it expanding use of Apple Health Records to embrace Allscripts. And in not unexpected news, AliveCor has dropped its KardiaBand accessory for the Apple Watch. AliveCor has been pushing into more specialized territory with its 6-lead ECG anyway. But does a 6-lead ECG take AliveCor away from the direct-to-consumer (D2C) model…? I think it might. Meanwhile, if you do see anyone with both hands on a device – while simultaneously pressing it to their left knee or ankle –  you’ll know exactly what they’re doing

Things seem to be quite peachy at not for profit hospitals, with Mayo growing profits by almost 300%.

Researchers are working on a biosensor that uses interstitial fluid and might be used in place of blood draws in the future.

Biobeat gets FDA clearance for cuff-less non-invasive blood pressure. This has potential. First of all, cuff-less non-invasive BP should mean it’s more comfortable for the patient. That could be especially great in the home for chronic conditions. Second, in my experience there aren’t many wireless BP devices cleared for use in the hospital. There are two unknowns for me that are crucial to success down the road. First, does it fit easily into the clinicians workflow? If not, that’s going to make adoption in the hospital an uphill battle. And second, do the economics work? If it’s more expensive overall than existing approaches, it’s probably not going to fly, however much more comfortable it may be for patients.

In the land of mobile appointment bookings, doc’s choke on Zocdoc’s new pricing model.

I think Remedy Partners really needs this merger with Signify Health. Remedy Partners is arguably the most important player in Medicare bundled payment pilots. It “owns” more episodes of care than any other entity, and in some cases took on the financial risk instead of the providers. But, most of the cost variation in each bundle is driven by post-acute care (slide 16 onward). That is, what happens after the patient is discharged from the hospital. Without a good way to monitor patients post-discharge, that creates a huge financial risk for Remedy. This merger with Signify Health can help to fix that. 

Two chairs aims to take the stress out of finding a therapist.

Strong preventative healthcare is woefully lacking in the US. As Amy Brown of SpringBuk says, incentives for preventative care are not aligned. Honestly, I think there are limits to how much analytics can help with that. There are two massively fragmented, multi-trillion dollar industries here (payers and providers are two different industries, they just share a value chain). There’s a bigger piece here about preventative care that needs to come together in my head, I’ll try and get it written in the next week or two.

And last but not least, Isansys pilot shows that heart rate variability could predict 90 day mortality in patients with cirrhosis of the liver.

Fortnightly Healthtech Update #3

Google’s Deepmind shows promise in early detection of kidney failure…and immediate takes heat for using an almost all-male training data set. I can definitely understand people’s concerns with that. On the other hand, I’m not sure there’s a deliberate gender bias here. It’s a machine learning project. Start with the data you can easily get, then refine with better data later. If you wait for the perfect data set, you might never get started.

This might be huge, CMS commits to puff up remote patient monitoring. CMS introduced new reimbursements for remote patient monitoring at the start of 2019. But astonishingly, managed to overlook the need for a precise definition of what services and devices would actually be reimbursable. Bit of an oversight that. This is going to get fixed for 2020.  Also stepping up to the plate is the granite state, providing Medicaid coverage for telehealth and remote patient monitoring. What a great way to provide better care for a rural population that is often under served because of the low reimbursement rates. As if to make my case, here’s another example of using telehealth to bring badly needed help to the patient, instead of forcing the patient to travel. And legislators are increasingly seeing telehealth as a way to help rural moms. Here’s an example of telehealth for preemies and newborns that need special care

Those new remote monitoring CPT codes really are catnip aren’t they…? New to me DynoSense, jumps in with FDA clearance for…I’m not quite sure what actually. The press release isn’t clear, but then press releases often aren’t. In the promo video, it looks like a direct to consumer offer. In any case, the sensor appears to yield lots of useful data for not much effort on the part of the patient. And that is going to be one of the keys for successful remote monitoring of chronic conditions.

Voluntis gets FDA clearance for app to help cancer patients self-manage their daily life. Interesting how in this deeper look, Romain Marmot the COO, says that oncology is basically a chronic disease in its own right that needs to be managed better. I like that perspective, thinking about one leg of the quadruple aim that perhaps does not get as much love as the others. And, I wonder how many providers think about quality of life for cancer patients when they are living day to day, outside the treatment center…?

In a somewhat similar direction, this article discusses how smart homes can become the epicenter for self-care. I mostly like where this is going, but only up to a point. I could take issue with phrases like “In the near future, smart homes will be primary care.” No, not in the near future on any significant scale in the US. There are 3 ways this scenario can become a reality: First, the widespread adoption of ACO-type payment models that shift the focus onto preventative care. Second, a large number of people go down the direct primary care path – and I think that’s really only an option for the reasonably well heeled. Third, primary care delivered via the smart home could take off if there’s a reimbursement code for it. And that means alignment between telehealth and/or remote monitoring billing codes and smart home solutions. In the rest of the world, it’s a different story. Where single payer systems dominate, preventative care has a much stronger focus. I can see much faster adoption of smart homes being used for healthcare in Western Europe than in the US. I’d love to be proved wrong on that, but I really can’t see it. And let’s be fair, Alexa’s latest skills are a long way from actually delivering primary care in the home. Those skills are really just nibbling around the edges of the healthcare problem.

OK, now I’ve flirted with the murky subject of the broken US healthcare system, we might as well get this out of the way: 40% of patients face surprise out of network bills. Many of those unpleasant surprises come from bills for ambulances. As that second article notes, many ambulance providers are out-of-network “because they couldn’t agree on a fair rate” with the health insurance companies. A more cynical person might feel that they were deliberately out of network so they could bill large sums to people who are at their most vulnerable and in no position to argue. Especially as that article also notes that ambulance services in the US used to be a public service, yet are now increasingly a for-profit enterprise. The whole idea that a company can provide services to an individual without their explicit consent and then just bill them whatever they want….that just doesn’t feel right now, does it.

When if comes to “wireless”, it’s all relative I guess…Helen DeVos Children’s Hospital in Michigan introduces a wireless solution for monitoring epilepsy. Not as wireless as the devices that I’m used to working with, but if it gets you down the path to better management of epilepsy, all power to you.

On the topic of wireless devices, I promised a deeper piece on iRhythm’s Q2 results and the 12 words that change everything for iRhythm and similar holter monitor competitors, such as BardyDX.

In patient monitoring, over-alarming is an ongoing problem, with 63% of alarms in the ER going unattended. I fully anticipate that machine learning is going to bring a huge improvement into the quality of patient monitoring in the next few years. Startups like PhysIQ are pushing down this path.  And HBR has a review of Current Health being piloted in remote patient monitoring – in the UK, so no reimbursement code change required to incentivize the change.