Contemplating the complexity and wonder of the human body and its systems can be mind-blowing, especially in the context of today’s medical science and diagnostic techniques. Many of us have been touched by some disease that requires numerous trial-and-error tests to identify our affliction.
The challenge for the medical community is that enormous amounts of digital and analog data are generated by instruments, images and tests. While powerful analytical tools and software can aid the medical practitioner, there is another vast and equally important data type required if a proper diagnosis and treatment are to be rendered.
These data types are textual and come in many different forms, including patient histories, doctors’ notes, medical records, insurance documents, medical research, journals, clinical trials, treatment therapies and regulations that are published daily. It is impossible for the best medical minds to process, synthesize and utilize this dynamically changing sea of textual information in combination with digital and analog data. Important insights and potential breakthroughs that exist between numerical data and textual data have remained unknown to the medical community until recently.
Over the last three years, IBM’s cognitive platform, Watson, has taken on this challenge, with promising results in oncology for the fight against cancer. In partnership with Memorial Sloan Kettering Institute, Watson was trained in oncology and showcased its unique capability to ingest and analyze a patient’s medical records to help the clinician identify evidence-based treatment options. Watson’s artificial intelligence (AI) functions understand, reason, learn and provide context for its analysis. Doctors interact with Watson throughout the process, verifying information and requesting other relevant sources of data to understand the holistic picture of the patient. Finally, Watson completes its analysis, identifies a priority list of treatment options based on Memorial Sloan Kettering Cancer Center’s expertise, and provides links for supporting evidence. However, Watson is playing purely an advisory role.
Now imagine this cognitive capability being applied to diagnose and treat a complex system such as an aircraft. As it is for a human, the goal is to remain “healthy” and in service for as long as possible. The typical commercial aircraft has critical systems and subsystems that require analysis, prognostics and diagnosis to determine its overall health and viability. The aircraft and its systems are instrumented to provide real-time data from sensors that report on the state of its operating condition.
An aircraft also has a rich history of textual information such as maintenance logs and records, QAR notices, manuals, regulatory alerts, configuration manuals, customization manuals, maintenance procedures, repair logs and parts notices. These textual data types all need to be read, understood and processed by those “doctors of maintenance” responsible for the lives of those flying on the airplane. But currently this textual data remains “dark” and not fully utilized. This deep reservoir of textual data is not being correlated with the real-time digital data produced by the systems on the aircraft. What important insights are being missed? Since this cognitive capability can detect sentiment and draw inferences from the language it learns and understands, what early warnings are being missed from this textual data lake? Are the digital and analog data telling you one thing and the textual data validating or refuting your conclusions and actions?
Today’s aerospace and defense industry is clearly focused on the promise of big data and analytics whose source is only digital and analog data types being acquired by sensors, black boxes and other data-acquisition devices. Data from sensors is clearly important but tells only part of the story.
Airlines, OEMs and MROs are confronting multiple issues in this transformation process:
- The modern aircraft collects more than a terabyte of data per flight, and the IT infrastructure for handling this torrent is seriously inadequate.
- The result is that much of the data is purged—that is, lost.
- Also, is the data from the aircraft systems being transmitted in a fast and secure manner?
- Are the turnaround times of the analytical/prognostic evaluations fast enough to catch serious problems before they become catastrophic?
- What resources and processes are being used to perform analytical diagnoses on the textual data set that characterizes the aircraft patient?
At IBM, we believe the aerospace and defense industry could benefit from the adoption of cognitive capability, just as the health care industry has. Several industrial and A&D companies have started applying these cognitive capabilities. We are committed to investing and maturing our cognitive/Internet of Things platforms, making them available to all industries and supporting them as a robust cloud service with a rich application program interface that will enable one of the greatest digital-transformation opportunities today.
Glenn V. Ries is a senior account partner for IBM Global Business Services with over 25 years in the A&D industry and extensive analytic predictive modeling experience.
The views expressed are not necessarily shared by Aviation Week.