Going Mobile-Tech requires Going Low-Tech: Lessons from an American Nuclear Submarine

Damjan Denoble

At the dawn of the microprocessor in the late 1970s, the U.S. military recognized that portable computing technology could potentially empower “field medics” to diagnose and treat a broader range of ailments.

A technological breakthrough of this type would have tremendous tactical value for missions that necessitated personnel are away from “home base” for a prolonged period of time. Any sort of injury or illness had the potential to interrupt or unwillingly sabotage a mission, all at great cost.

The NAVY’s nuclear submarines probably stood to gain most from such a breakthrough. Submarine missions required that crews stay submerged for months at a time, with no doctors or specialists on board (and if submarine movies are any indication, missions required a minimal amount of moaning from crew members since enemy subs could detect even the tiniest noises in surrounding water). Instead, all medical duties were assigned to the “Independent Duty corpsman.”

To ensure the healthiest crews possible, NAVY submarine recruits were put through a series of extensive health checks. Consequently, odds of acute conditions requiring immediate surgery occurring onboard the submarines were small. Nevertheless, due to the relatively high incidence of acute abdominal ailments that could mimic appendicitis, the risk of misdiagnosis and unnecessary interruption of operations was significant enough to warrant the search for a solution.

The solution that the NAVY tested and eventually adopted was an MS-DOS support algorithm for submarine medics aptly named “The Abdominal Pain Medical Support Program,” and it could run on a Z-100 PC Series Computer. The computer program algorithm was designed using a BAESIAN approach, which determines the increase or decrease in the probability of a suspected diagnosis based on previous knowledge about the nature of the relevant disease, its prevalence in a defined population, and all additional information collected in diagnostic processes to which the patient is submitted.

After years of development and calibration the support algorithm achieved a success rate closed to that of an expert physician involved in a hands-on diagnostic process. It was the Independent Duty corpsman’s duty to use his knowledge of signs and symptoms that characterize appendicitis, to substantiate the diagnosis.

But, because the effectiveness of a BAESIAN analysis depends on the availability of “a priori knowledge,” the history of the research leading up to the Abdominal Pain Medical Support Program suggests that the NAVY struggled to create an effective computer algorithm, even though it benefited from the advantages of a mandatory data collection system and a homogenized patient profile.

Indeed, the NAVY thought computer-assisted diagnosis so daunting a task that the MS-DOS support model was never tested outside of the submarine setting, since it was thought that the success rate would probably compromised by many confounding ailments that could present symptoms similar to appendicitis.

Within the confines of this struggle lies the lesson for governments and other organizations looking to implement mobile health technology to extend the reach of disease diagnosis in resource-poor regions;

The enormous data-collection challenge faced by health and medical professionals must be addressed before the introduction of the technology itself.

If guaranteed and unfettered access to patients did not guarantee the NAVY a sufficient base of a priori knowledge to support an algorithm designed to help diagnose a single ailment, we can safely assume that the challenges of a project officer working with a patient population on the outskirts of the Karoo are going to be much greater.

By the same token, we should not assume that the era of evidence-based medicine has done away with the problems of symptom and history-based diagnosis. Rather, precisely because modern medicine is based on high technology and sophisticated biological and chemical tests, it is both native and disingenuous to assume that poorest people in the world’s most resource-poor regions will soon enjoy the benefits of a comprehensive medical history based on laboratory work.

For mobile health diagnosis in resource-poor countries to be successful, therefore, low-tech diagnostic skills may have to be reestablished as a core component of the modern physician or health workers skill set. This includes a return to the traditional, painstaking process of querying and observing patients in a search for signs and symptoms; a process which so-called “objective” diagnostic methods often bypass.

This will take time. If it took years to figure out how to use a mobile interface to diagnose a submarine crew, we can anticipate that the process will be more tedious when it involves entire continents. Fortunately, it does not mean that mobile health care is impractical in the fight against undiagnosed disease. In the case of some diseases with a global impact like malaria, the diagnosis may still be established based on signs and symptoms alone.

Bottom line, however, whatever disease be the subject of a mobile healthcare endeavor, it will be important to lead with the necessary low-tech process and not with high-tech assumptions.