Better quality medical care: There's an app for thatâ or there will be soon
When Dottie Spencer was released from the hospital after a minor surgical procedure, she felt she was doing great, and the sensor package she’d just swallowed agreed. The pill-sized ingestible device transmitted temperature, heart rate, glucose levels, and hormone balance readings that all were within the normal range. She had dinner with her husband and went to bed.
Spencer was sleeping soundly that night when a cloud-based machine learning algorithm detected a potential anomaly in the data streaming through an app in Spencer’s smartphone. Based on the accumulated data of millions of other patients who had used ingestible sensors like hers, the algorithm recognized an elevated probability that Spencer was experiencing the earliest stages of sepsis. The system texted Spencer’s doctor, who approved its proposed course of action. An automated system telephoned Spencer, alerted her to the issue, and sent a car service to take her to a nearby urgent care facility, where her doctor examined her through a telemedicine link. Spencer got a prescription and the car service took her home.
Spencer later told her husband that if the AI hadn’t told her there was a problem, she’d never have guessed she wasn’t in tip-top shape, but it sure was nice to know that someone—or something—was keeping an eye on her.
Recalibrating the time machine
An admission: Spencer’s story didn’t take place in today’s healthcare system. The technology described won’t be available to patients for the next five to 10 years at the earliest. But it’s under development now, and engineers, doctors, and healthcare providers are preparing for a major shift in how the healthcare industry serves patients. In short, it’s about using networked sensors to collect vastly greater amounts of data about patients and using distributed computing to deliver optimal treatment.
“We are digitally transforming analog bodies,” says Rich Bird, Hewlett Packard Enterprise’s worldwide industry marketing manager for healthcare and life sciences. “We are taking people and turning their human physiology into digital signals that we then use to deliver healthcare, which is more efficient and convenient for the patient and which makes the clinician’s job easier so that they can focus on spending more quality time with those patients.”
To achieve this revolution will require technological advances across a series of related fronts. One is the sensors that collect patient data.
Already, most patients carry smartphones that can measure a surprisingly large variety of information about their health, both mental and physical. For instance, Mindstrong Health, a California start-up founded by three physicians, has developed an app that can continuously monitor users for signs of depression just by how they type, swipe, and scroll.
Meanwhile, smartphones are serving as wireless networking hubs for other wearable medical sensors, such as devices that measure temperature, heart rate, and oxygen level. In the future, these sensors will improve. And they’ll be joined by new kinds of devices that are as yet only under development, such as ingestible sensors that are able to detect the presence of dangerous bacteria or cancerous cells.
“Digital transformation is about patients caring for themselves,” says Bird, “whether that be by wearing Apple Watches, Fitbits, using apps on their phone, or letting sensors send their biological readings into an app.”
The Cleveland Clinic, an international network of hospitals and health centers headquartered in Ohio, anticipates the development of sensors that will be permanently implanted into patients’ bodies. “Pressure sensors in the heart and in the pulmonary artery, for instance, will transmit via a smartphone to a central monitoring unit,” says Dr. Peter Rasmussen, the Cleveland Clinic’s medical director of digital health. In essence, every person’s body will be dotted inside and out with tiny weather stations reporting on all manner of vital signs.
Some devices will do more than just sense. For patients with Parkinson’s disease, implanted brain stimulators can ease symptoms. In the future, they will “not only stimulate but also record how the tremor’s responding to stimulation,” Rasmussen says. If the data warrants it, a doctor could opt to change the settings on the simulator without even having to see the patient in person. Similar approaches might someday be applied to epilepsy, depression, pain management, and a host of psychiatric and neurological conditions.
Remote assessment and treatment will help pave the way for the “hospital everywhere,” in which technology provides much of the benefits of inpatient care without the attendant cost and inconvenience. “There’s pressure on hospitals to get patients out of the hospital faster and sooner,” Rasmussen explains. “If you can shave one or two days off of a lengthy hospitalization, finish up the last couple of days of management at home, equipping with them some type of sensor package, there’s a lot of promise for cost savings for that. And it’s obviously better for patients to be home again.”
The challenge of data storage
The digital transformation of healthcare is going to require a lot of data, of course, and it’s going to have to be stored somewhere. Legacy databases aren’t up to the task. Electronic health records, or EHRs, have been kept since the 1970s but haven’t evolved to keep up with today’s flood of data. At the moment, says Rasmussen, “there are not great ways to move large amounts of data from our EHR into the hands of the patient.”
There have been small moves to improve the situation. For instance, Epic, the largest health records software company in the United States, updated its popular MyChart app to allow users to pull data from Apple’s Health app. But it’s not yet possible to move patient data from Epic to other major EHRs.
To meet future needs, healthcare records will move to an IP-based infrastructure that feeds data to AI algorithms, all while securing the data against theft or tampering.
“To revolutionize health, we need data,” says HPE’s Bird. “First, to train the AI algorithms, and then to run the analytics on so we can predict whether a patient is becoming more ill.”
The U.S. Food and Drug Administration is already getting ready for the AI future. Earlier this year, the FDA approved its first AI-based product, IDx-DR, a device that analyzes retinal images to detect diabetic retinopathy. More AI applications are sure to come down the pike with increasing frequency.
Patients aren’t going to be the only ones benefitting from medical AI—physicians will, too, and not just in terms of diagnosing ailments. While in the past many physicians have grumbled that EHRs actually increased their workload, Rasmussen believes that in the future, intelligent systems will serve as electronic amanuenses, monitoring patient visits and harvesting information passively from the environment. “Maybe the microphone in a computer will pick up the conversation between the physician and the patient,” Rasmussen says. “Then the AI will remotely extract the relevant information and package it up into notes for the physician to review and sign off on.” Such a system could be linked to cameras in the examination room that feed back information about a patient’s body language and facial expressions. “An AI system could glean insights into comprehension and anxiety states and fear, and that could be fed back to the caregiver to coach them in how to interact with the patient,” Rasmussen says, adding, “Physicians are not necessarily [good] at human interaction.”
If eavesdropping AIs and internal weather stations sound like the stuff of science fiction, that’s only because technology has truly brought healthcare to the cusp of a breathtaking inflection point. “In my mind, the practice of medicine has not seen this big of a cultural shift since the implementation of electronic health records half a century ago,” Rasmussen says. “And really, we’re still in preschool in terms of the state of development. We have a long way to go.”
The future of healthcare: Lessons for leaders
- In the next five to 10 years, a new generation of small networked sensors will provide clinicians with up-to-the-moment, multifaceted insight into patients’ health.
- Integrating masses of medical data into a comprehensive picture of each individual’s health will require both advanced data architecture and sophisticated AI.
- Next-generation machine learning algorithms will not only keep a watchful eye on patients’ health, but also direct and even provide treatment.
This article/content was written by the individual writer identified and does not necessarily reflect the view of Hewlett Packard Enterprise Company.