This Aviation Week op-ed describes efforts to advance aviation data analytics. Embry-Riddle Aeronautical University President P. Barry Butler, Ph.D., writes that the rapidly growing field has the potential to save lives and reduce accidents. This essay — published in October 2022 — is available to read at the Aviation Week website and below. 

By P. Barry Butler

Advancing aviation safety is a deeply personal goal for Marisa Aguiar, who earned her Ph.D. in Aviation from Embry-Riddle Aeronautical University last year. 

Her father, Simmons Airlines Capt. Orlando Aguiar, lost his life along with 67 other people in 1994, when American Eagle Flight 4184 crashed in Indiana. After being instructed to hold at altitude amid freezing rain, Aguiar and First Officer Jeffrey Gagliano struggled mightily to overcome icing that had built up on their twin-turboprop ATR 72-212. Both were highly capable aviators, but the icing was severe and the aircraft went down. The tragedy prompted multiple FAA Airworthiness Directives on deicing procedures and requirements for certain turboprop aircraft.

To help pilots survive complex in-flight emergencies such as the one her father faced, Marisa Aguiar’s doctoral dissertation sought to advance aviation data analytics — a rapidly growing field aimed at saving lives and preventing accidents.

Her project used a Safety Performance Index algorithm to flight and maintenance data patterns. The algorithm examined how different decisions might reduce risks within a large Title 14 CFR Part 141 collegiate flight-training operation. For instance, how would hiring more flight instructors improve operational safety? How does safety correlate with the number of maintenance technicians, flight students and aircraft in the fleet?

Researchers such as Aguiar, who now teaches for Purdue Global, are applying aviation data analytic advancements rapidly.

At Embry-Riddle, an avionics system in the university’s training aircraft and simulators automatically sends flight data to safety servers. The technology allows us to spot excess speeds, pilot errors, maintenance problems or other issues. By year-end, we aim to take our analytics to the next level, by setting up a flight safety dashboard. Initially, the dashboard will allow us to combine flight, scheduling, maintenance and other data so that we can more easily visualize risks and measure safety performance, notes Dr. Alan Stolzer, dean of our College of Aviation at Daytona Beach, Florida. We are also using these tools in our GE Digital Safety Lab for educational purposes and to prepare future workforce, too.

Our goal university-wide will be to leverage predictive analytics. In particular, we aspire to predict the likelihood of aviation incidents, such as unstable approaches, runway incursions and loss of control. “If we could develop an algorithm to do that,” explains Dr. Dothang Truong, program coordinator for the aviation Ph.D. program, “the safety officer could see not only the visualizations of data patterns but also the probability that some incident might happen. That prediction would allow them to develop a strategy to prevent it from happening.”

As an example of predictive analytics, Edwin Odisho, another recent graduate with a doctorate in aviation and now an American Airlines pilot, analyzed four years of data collected by NASA from 35 regional jets to build machine-learning algorithms capable of predicting the likelihood of runway excursions. The ability to predict pilots’ misperceptions of runway excursion risks could contribute to more effective simulator training scenarios and strategies, Odisho noted. The information could also be incorporated into avionic display tools to alert pilots to danger.

For researchers in this field, a key challenge is that airlines must deidentify sensitive safety information before sharing it, which takes time. Progress is being made, however. For example, through a public-private collaboration known as Aviation Safety Information Analysis and Sharing, spearheaded by the FAA and facilitated by MITRE Corp., airlines and some business aviation and charter operators routinely share deidentified safety data amongst themselves and the FAA. By aggregating safety data, they can learn when seemingly “one-off” events are part of a more widespread problem.

On the horizon, Truong says, it should become increasingly feasible to analyze both quantitative and qualitative information to assess human factors. Natural-language processing, aided by machine learning, might even allow us to predict risks related to stress, fatigue or depression.

At our Center for Aviation and Aerospace Safety at Embry-Riddle, we work to squeeze every possible lesson from aviation safety data. The center, directed by Robert Sumwalt, former head of the NTSB, was established to improve safety for all who fly. The U.S. airline accident record over the past decade has been outstanding, Sumwalt says, yet smaller safety-related events happen daily and require scrutiny. Data analytics programs “will prove essential to helping U.S. carriers continue the quest for safety,” he says, “and they will benefit global air carriers to improve their safety record, also.”

By advancing aviation analytics, our objective is to reduce aviation risks for everyone, in honor of Marisa’s father and others gone too soon.