Covid-19 is driving a telehealth revolution — more diagnosis and treatment going 'mobile'
From blood pressure monitoring to glucometers, health apps on our phones have been changing the health care scene for more than a decade now. But the recent global pandemic might be able to push the boundaries of what the smartphone can do. In an online presentation organized by the AI for Good Global Summit, Shwetak Patel, computer science professor and director of Health Technologies at Google, describes how we can reimagine the potential of our daily handheld devices. The recording and presentation are available here.
Why is this important? According to Patel, one of the questions that are at the top of agendas for researchers today is: “How do you predict a disease if you’re asymptomatic (or before you’re symptomatic)?” In the past months, Covid-19 has revealed the weaknesses of healthcare systems around the world, including the overburdened capacity of hospitals and laboratories. To help alleviate this strain, computer scientists have been rethinking how the “most ubiquitous” platform in our life — the smartphone — can be used as a relatively inexpensive healthcare tool, beyond its well-documented ability to conduct contact-tracing.
Bringing healthcare into our personal space. While the last decade has seen the increased use of mobile devices to enable faster diagnosis and more efficient delivery of everything from polio vaccines to HIV prevention, the coronavirus pandemic is further accelerating the paradigm shift. According to Patel:
“Telemedicine and remote care have been expanding rapidly over the past few years; you don’t need to go to the clinic anymore to administer care.”
As patterns and biomarkers emerge from available data regarding Covid-19, mobile technologies can contribute to the diagnosis, treatment, and assessment at both the personal and policy levels:
“Mobile devices, because they are so personal, are an opportunity to drive screening… based on what the phone has,” Patel adds. “You don’t have to be perfect. You just need to have insight for what to do next… enough confidence to do a triage on a person.”
Same phone, new tricks. By leveraging the sensors already present in smartphones, such as microphones, cameras, and accelerometers, basic screening and monitoring for common disease biomarkers can be scaled up. Here are some of the technologies implemented or currently being developed by the Ubiquitous Computing Lab at the University of Washington:
SpiroSmart. Home spirometers — a device that measures how much air is being exhaled — are used more and more to monitor lung function and evaluate possible breathing problems. However, as they’re either too big or too expensive, this app is a low-cost alternative that performs spirometry by asking an individual to breathe into the built-in microphone on his or her cellphone. The app records the data on the device, and you can share it with the physician, who can then screen and diagnose varying degrees of obstructive lung ailments.
BiliCam. By using the built-in camera, this mobile app is able to monitor bilirubin levels in the blood—a necessary measurement to detect the possibility of newborn jaundice. As bilirubin levels peak 4-6 days after a baby is born (i.e., once the baby has left the hospital), the app provides a more accurate way of detecting jaundice that is difficult to determine by eye.
HemaApp. The app provides a low-cost, non-invasive alternative to screen blood for hemoglobin levels, using a phone’s built-in camera. With the use of a light source shining through a finger, the app does a chromatic analysis, analyzing the color of a patient’s blood to help screen anemia and assess a patient's response to iron supplement treatments.
Issues in scale-up. While these are promising developments in the healthcare and AI fields, Patel reminds us that there are other considerations involved in putting such apps into widespread use. Regulation of these new apps as healthcare devices by national and regional authorities would ensure that such apps will be of standard quality. In the absence of regulation, however, such apps are often perceived as inaccurate, even if they’re already comparable to available medical devices.
Issues of privacy also abound. There may be unintended consequences, should the data be used in a different way than originally intended. One, oft used example is health insurance providers - who could potentially use such data to make judgments about people with pre-existing conditions, without the knowledge or permission of the users.
Finally, one still needs to take into account other factors such as where one lives, their environment, and their existing support structures, Patel says. Although diagnostic apps can help advance screening of health conditions, solutions also need to follow — particularly to disadvantaged groups.