Digital health conferences, especially those held in the Bay Area, tend to overemphasize the power of technology in solving all that ails health care.
At the two-day Rock Health Summit in San Francisco that concluded Wednesday, the conversation, refreshingly, centered on the practical and down-to-earth as opposed to the ivory tower perspective of healthcare’s salvation by technology.
When Investment Rhymes with Canada
Canada has a proud history of achievement in the areas of science and technology, and the field of biomanufacturing and life sciences is no exception.
Here are some key takeaways gleaned from the main day of the conference.
Start with the problem, then think about technology
Omada Health is perhaps the one company that people evoke when they think of a successful digital health startup that has bridged the wide chasm between pilot and adoption. The San Francisco company seeks to change behaviors through a mobile app aimed at prediabetes patients to prevent them from getting the full-onset of that chronic and debilitating disease.
At the summit, the company’s CEO, Sean Duffy, was asked how the company understands not only digital health design but for whom it can work successfully. Duffy explained that it’s tempting to fall in love with the technology being developed.
“I love technology. I am such a gadget geek. I have an automated floss dispenser. You hit it and it shoots out the floss,” he said. “But I try really hard not to fall in love with the technology.”
Duffy explained that the best way to build a product is to talk with people that have a problem.
“You have to start by listening,” he said, adding that even before the product was launched, they visited people’s homes as far as Georgia to talk with people who were told they were prediabetic. “You have to have the marriage of the qual and quan because from the quantitative side you can see demographic differences in usage, you can see certain populations using certain areas of the program more. [But] bring your users in your company. Don’t forget about who’s actually using it. It’s easy to detach from the brain on the other side.”
Another panelist said some people have the “shiny object problem” and in a light-hearted dig at Duffy referred to the “dental floss” – something that’s cool but doesn’t really alleviate any unmet need.
“Those things just don’t stick cause it wasn’t solving a problem and you just force into an organization and so we always start our meeting every year with, ‘what are the key problems that we have to solve’ and then go and look for those solutions instead of responding to knocks on our door,’ said Veenu Aulakh, executive director, Center for Care Innovations, a nonprofit that funds efforts and helps facilitate collaborations to improve California’s safety net.
AI is the most overhyped/high-potential area in healthcare
Aulakh referred to VR and AI as shiny objects, but the latter technology is more complicated than that and two snap polls conducted with the audience proved that. AI is simultaneously the most overhyped area in healthcare and also the one that holds the most potential to transform it, per the audience.
This is how Dr. Jessica Mega, chief medical officer of Verily, the life sciences subsidiary of Alphabet described the hype.
“One of my very favorite colleagues at work – his pet peeve is when people say “We are going to take it and machine learn on it. We are going to AI it up,” Mega said to audience laughter.
In other words, the AI conversation in healthcare desperately needs a reality check.
Mega said one achievable application would me in medical image detection.
“In diabetic retinopathy …there is an application where you can take the images, train the images and come up with some meaningful insight,” she said noting that this would serve an unmet need. “The important thing is that you work with regulators to ensure that it is a safe and effective product.”
Earlier in the day, FDA’s associate director for digital health, Bakul Patel, said that AI has flipped fundamental science on its head.
“[With AI] I have boatloads of data. I don’t know what is important in here but once you run it through learning I will know what outcomes it gets. So it’s almost discovery and learning that yields a model that will help,” Patel said. “If you take radiological applications and highlighting a spot in the image – it can be done in many different ways. With machine learning …can we show the same efficacy and this performance of [detecting that spot] on that radiological image? Can we do that consistently? That’s what we care about.”
Back in January, FDA cleared the first deep learning and AI application in January paving the way for Arterys Cardio DL product. Radiologists have typically relied on software to manually segment and draw contours around the ventricles to determine how the heart is functioning. Arterys’ AI-assisted software automates that process thereby generating contours of the insides and outsides of the heart’s ventricles to speed up the segmenting and contouring.
Don’t underestimate the importance of Quality Management Systems
While AI Is overhyped but exciting, digital health companies need to stay on top of more humble technologies that can really make or break them. And that was the implication from an answer to a question from the audience regarding how digital health startups can prepare for FDA guidelines on digital health in the future.
“As a tech startup founder, one of the first things I would do is figure out what a QMS is – a Quality Management System and do it,” said Christine Lemke, president of Evidation Health, a startup that works to clinically validate and provide the economic evidence behind digital health products/services. “Institute a Quality Managment System immediately – [it should be] the first thing you should do.”
Lemke added that this should be done in advance of conversations with the FDA or providing data to the FDA because it’s more difficult to have to go back in and change processes.
“Do it now,” she urged.
Photo: Natali_Mis, Getty Images
