Using FHIR and AI to Expand the Reach of Healthcare

Alani Lee is a rural native of Hawaii. Her home is in Ni'ihau, a remote island over 150 miles away from the state’s capital city of Honolulu. However, it's this remote location that has instilled a deep sense of belonging amongst the island's residents. Alani is a regular contributor to this sense of goodwill as the operator and proprietor of the town's local ice cream parlor. Heavily admired within her community, locals visit her storefront for good conversation just as much as for good ice cream. Happy to provide both, Alani enjoys nothing more than being a positive presence to every customer who walks inside her store.

Unfortunately, this cherished engagement with her patrons was one day sadly disrupted after Alani was unexpectedly diagnosed with late-stage breast cancer. Her experience showcases the complications that arise when treating patients living in remote areas and how it's being solved through the collaborative efforts of Smile Digital Health and Red Hat.

Shortly after her diagnosis, Alani underwent a partial mastectomy to remove her tumor. While the surgery was a success, the late-stage recognition of her cancer led to it spreading beyond her breast. As a result, Alani developed metastatic brain lesions that required a combination of corticosteroids and radiation therapy to treat. While she was now cancer-free, her immune system was severely compromised after withstanding the rigor of invasive surgery and extensive steroids exposure. Alani faced a long road to recovery before she could hope to feel herself again. Her physician's main concern was Alani's susceptibility to sepsis, a dangerous condition that had the potential to affect her well-being even further.

Understanding the Complexities of Sepsis

Sepsis is a condition where the body's infection-fighting response becomes overactive and, as a result, causes damage to the surrounding tissue. The survival rate of sepsis varies, but its most critical mortality factor is based on how early the condition is recognized. More specifically, every hour sepsis isn't detected, a patient's chance of dying rises by 7.6%.

Once diagnosed, the primary treatment for sepsis is antibiotics; having said that, the most effective way to avoid serious complications is through early detection. While there is no single diagnostic to detect sepsis, its insidious emergence can be recognized through any combination of the following symptoms:

Extreme pain or discomfort
High heart rate
Fever, shivering or feeling very cold
Confusion or disorientation
Shortness of breath
Clammy or sweaty skin.

No Place Like Home: Choosing the Optimal Location for Recovery

When it came to her recovery, Alani was left with two choices. She could opt for long-term residency to ensure that any abnormalities in her vitals could be monitored for sepsis, or return home in hopes that her improved quality of life amongst family and friends would aid in her body’s recovery. Unwilling to spend countless weeks away from her community, Alani chose to leave the hospital and return home to recuperate amongst her loved ones.

Unfortunately, returning home meant putting over 150 miles between Alani and her care team. In the past, a decision like this came with a significant risk. Outpatients in remote locations often had to live with the fact that a physician wouldn’t be close at hand to address any complications during their recovery.

Smile, Red Hat and a collection of other collaborators recognized this limited reach in healthcare and created a partner-driven solution that could deliver care anywhere, regardless of location.

Advanced Remote Monitoring Through FHIR and AI

Smile and Red Hat's collaborative efforts resulted in an innovative remote diagnostic tool. Recently developed, it possesses the theoretical functionality to use AI and Event-Driven Architecture to detect various forms of decline in a patient's health. Monitoring is achieved by connecting and streaming data from various medical devices used by patients when living remote from their care team. Using Red Hat’s OpenShift Container Platform, deployment and management of novel Machine Learning Models (AI/ML) to remote locations is now possible, allowing for active monitoring of emergent conditions like sepsis, heart attacks, strokes and pulmonary embolisms.

In Alani’s case, the tool would be deployed in her local clinic and used to monitor for the condition of sepsis. Her AppleWatch, EHR, BloodPressure Monitor and At Home Questionnaire data would all be connected to the diagnostic tool. The information from these devices would then be intelligently ingested using the interoperable capabilities of Smile to provide a readable pool of data under the (Fast Healthcare Interoperability Resource) HL7® FHIR® data standard that could be assessed for risk by Red Hat’s AI platform.

Any change in data that signaled an emergence of sepsis would be flagged and promptly communicated to Alani's care team. This digital connectivity between Alani and her physician would mitigate the limitations of their geographical distance, allowing for the intimate monitoring practices needed to potentially save her life.

The Power of Connection: How the Remote Diagnostic Tool Works

Prior to exploring what Alani’s experience would be like with the diagnostic tool, it’s worth examining how the tool's system of engagement functions during an emergency situation. In short, the data flow is designed to read, assess, flag and send any potential health complications to the hospital. Any verifiable risk will be expressed through a task sent to the primary physician’s computer. Depending on the results, the physician can send care directly to the patient, order their transport to the hospital or disregard medical intervention and continue home monitoring.

If the task isn’t tended to in a timely fashion, the prompt will be redirected to the hospital’s on-call doctor. This system of engagement allows for life-saving medical responses to critical health complications, as physicians are empowered to focus on patients before their most severe symptoms present themselves.

Since the diagnostic tool learns and improves by ingesting a holistic collection of patient data, Smile was the perfect partner due to its robust data integration platform that allows patients to utilize multiple devices to monitor their risks. More devices translate to more data, which improves assessments by increased speed and accuracy. Without this convenience, patients and physicians would be forced to select a single data standard to collect pertinent medical information, a restriction that would diminish the efficacy of the diagnostic tool. Instead, FHIR empowers physicians to choose any collection of data sources that can best guarantee proactive medical intervention.

Just in Time: The Early Detection of Alani’s Sepsis

Upon returning home, Alani felt like her body was on the mend. Each day she found a little more strength and this progress made her optimistic for her recovery. Driven by the desire to get back to work and her customers, Alani diligently followed her physician's rehabilitation plan. Every day she participated in gentle exercise, prioritized bed rest, took her medication and adhered to a simple and clean diet.

Alani's progress was tracked through a network of monitoring devices, including her Apple Watch, Blood Pressure Monitor and At Home Questionnaire (sent to her physician daily). An aggregated pool of data was then made interoperable through FHIR and assessed for risk by the Sepsis Detection ML model on OpenShift.

One Wednesday, Alani woke up with the slight sensation of back pain. While its intensity was trivial, she decided to mention it in her at-home questionnaire. The next day Alani's back pain intensified, and by the evening, she noticed an unusual amount of perspiration being produced from her palms. In an attempt to ease her worries, Alani checked her blood pressure hoping that the results would reassure her everything was fine. Instead, she experienced the opposite as she watched her blood pressure register much higher than normal. Alani resigned herself to the fact that she was most likely experiencing complications in her recovery. She decided that if her symptoms continued to evolve, she would schedule a trip to the doctor on Friday as she didn't want to trouble her son while he worked during the week.

Around 2:00 AM that night, Alani received a call from her doctor informing her that a helicopter was being sent to take her to the hospital. While on the phone, she noticed a significant increase in discomfort compared to when she had previously gone to bed. Responding to the urgency in the doctor's voice, Alani called her son, who helped her get prepared for the helicopter's arrival. Upon her arrival to the hospital, it was confirmed that she had entered into an early state of sepsis and was immediately treated with antibiotics.

FHIR-Enabled Timely Treatment

Rewind to Wednesday evening, when Alani was wrestling with the potential severity of her new symptoms, the diagnostic tool was collecting valuable data from her at-home devices. Her AppleWatch tracked an increase in Alani's body temperature while her blood pressure monitor registered a rise in blood pressure. These two data standards, combined with the information from Alani's at-home questionnaire, were aggregated through the interoperable functionality of Smile and provided enough information for the diagnostic tool to contact a doctor.

Sepsis is a time-sensitive condition where every hour counts. Alani's situation exhibits the importance of FHIR, as it allows for a comprehensive examination of medical information collected from many data points that can be the difference in saving a life. Had Alani relied on just one of the devices, medical intervention would have arrived too late.

Luckily for Alani, her early detection of sepsis resulted in a brief stay at the hospital, where she was able to quickly fight off the infection with antibiotics before her symptoms became severe. Upon returning home, Alani felt comforted knowing that, despite her care team's geographical distance, their focus and attention were always on her due to the connective power of the remote diagnostic tool. As a result, she continued to put all her energy into her recovery and was back happily managing her ice cream parlor a few months later.

Alani’s story highlights the importance of Smile and Red Hat’s collaborative efforts. When FHIR empowers intelligent platforms like OpenShift, healthcare can achieve faster data-driven outcomes and improved care that directly saves lives. In the case of Alani, it was her provider’s willingness to embrace the FHIR standard that enabled the remote monitoring needed to detect Alani’s sepsis. Providers keen on standing at the forefront of innovation need FHIR to champion new software capable of expanding the reach of healthcare across the globe. Smile provides our clients with the platform and services to facilitate these advancements and guarantee life-saving healthcare for patients like Alani Lee.