Empowering our healthcare warriors with predictive analytics

What can we, at Falkonry, do for the healthcare warriors who are at the front lines of the COVID-19 battle?

Our search started with this simple question.

The pandemic continues to spread rapidly around the world, infecting more than 1.9 million people across 213 countries (as of April 16, World Health Organization). With new cases arriving at overwhelming speed and scale, healthcare professionals are struggling against forces out of their control: overcrowded hospitals, shortage of ventilators, protective gear and other essential supplies, lack of coherent data, and the risk of getting infected themselves. Simply put, the number of infected people seeking care far exceeds the capacity of our healthcare systems.

At Falkonry, we started looking at ways we can use predictive analytics in healthcare to empower our clinicians and maximize the throughput of patients through the care process as well as improve the quality of care. We could improve care effectiveness by computer-assisted interpretation of real-time device and physiological information in the context of a patient, enabling clinicians to make rapid, effective decisions. This would essentially widen the “umbrella of care” that a single clinician can provide and hence directly impact ICU throughput and safety.

Current state of emergency medical treatment

Emergency medical treatment is carried out with class II devices assisted by physiological monitoring and labs. However, this data is vast and the usage of predictive analytics in healthcare still limited, making it a challenge to combine all the data into an effective decision support system. The largest proportion of emergency medical treatment among adults is necessitated by lung infection. Mechanical ventilators are a key tool used to manage treatment of such patients. There are many manufacturers of ventilators in the market today; from new and just delivered, to ones from storage, to loaners from colleges, universities, and nursing tech schools.  This heterogeneity creates compatibility problems with existing hospital devices when it comes to data exchange, not just between devices, but on the Hospital Information Systems (HMS) as well.

Ventilators, vents, operate at various modes, which vary between manufacturers.  Even with years of vent experience, clinicians struggle to find the proper settings from patient to patient.  During a patient’s stay, a clinician may find adjustments are needed based on visual analysis of other devices connected to the patient and/or verbal and physical feedback from the patient himself.  Vents provide in-the-moment and periodic snapshot information but no condition predictions.

Furthermore, while the news coverage of the COVID-19 emergency has focused on potential shortages of ventilators, the real shortage that cannot be dealt with by medical device manufacturers is the shortage of trained, skilled clinicians. It follows, therefore, that an urgent goal for predictive analytics in healthcare is to support increasingly fatigued healthcare warriors on the front lines by acting as a “force multiplier” during surges of critically ill patients in a pandemic.

Real-time sensor fusion AI solution

Falkonry proposes a Predictive Patient Monitoring solution targeted at improving COVID-19 patient outcomes for the most critically ill population who have to be put on ventilators. This approach to using predictive analytics in healthcare can help clinicians, especially intensivists, manage hospital surges and prepare for the coming waves. The solution includes both sensor fusion (integration of ventilator and other vital patient data) and real-time, autonomous machine learning that can discover patterns automatically and tune itself via clinician feedback in situ.

This combination solves critical problems in ICU care – lack of real-time, holistic decision support tools that take all available data into account, the trial-and-error nature of respiratory therapy for a novel disease, for which existing protocols are not sufficiently effective, and the relative lack of skilled, well-rested respiratory therapy clinicians in the midst of a pandemic surge.

The result is machine-assisted, real-time monitoring that pushes the boundaries of predictive analytics in healthcare by taking the entirety of the patient into account, providing actionable intelligence that increases each clinician’s ability to cover more patients even as the disease evolves.

We need to widen the umbrella of care that a single clinician can provide in order to overcome the pandemic. Next only to personal protective equipment (PPE), this is the most important preparation that is needed to manage COVID-19. Contact tracing will help us reduce exponential disease spread but expanding the umbrella of care will help us lower mortality.