DXplain

DXplain is a Clinical decision support system (CDSS) available through the World Wide Web that assists clinicians by generating stratified diagnoses based on user input of patient signs and symptoms, laboratory results, and other clinical findings ^[1]. Evidential support for each differential diagnosis is presented, along with recommended follow-up that may be conducted by the clinician to arrive at a more definitive diagnosis. The system also serves as a clinician reference with a searchable database of diseases and clinical manifestations.

Additional recommended knowledge

How to ensure accurate weighing results every day?

Guide to balance cleaning: 8 simple steps

How to quickly check pipettes?

History

Designed by the Laboratory of Computer Science at the Massachusetts General Hospital, work on DXplain began in 1984 with a first version being released in 1986 ^[2]. Today, DXplain contains a database with crude probabilities for over 4,900 clinical manifestations that are associated with over 2,200 unique diseases, yielding a total of over 230,000 unique finding-disease interconnections ^[2].

Educational Tool

Use of DXplain as a tool for medical consultation has been common to some institutions since it fills a gap, particularly for medical students in clinical rotations, that is not adequately covered by textbook literature ^[3]. The system's large and accessible knowledge base combined with its ability to formulate diagnostic hypotheses have made it a popular education tool for US-based medical schools; by 2005, DXplain was supporting more than 33,189 total users ^[4].

Methodology

DXplain generates ranked differential diagnoses using a pseudo-probabilistic algorithm (^[5]. Each clinical finding entered into DXplain is assessed by determining the importance of the finding and how strongly the finding supports a given diagnosis for each disease in the knowledge base. Using this criterion, DXplain generates ranked differential diagnoses with the most likely diseases yielding the lowest rank. Using stored information regarding each disease’s prevalence and significance, the system differentiates between common and rare diseases.

Accuracy

Analysis of accuracy has shown promise in DXplain and similar clinical decision support systems. In a preliminary trial investigation of 46 benchmark cases with a variety of diseases and clinical manifestations, the ranked differential diagnoses generated by DXplain were shown to be in alignment with a panel of five board-certified physicians ^[6]. In another study investigating how well decision support systems work at responding to a bioterrorism event, an evaluation of 103 consecutive internal medicine cases showed that Dxplain correctly identified the diagnosis in 73% of cases, with the correct diagnosis averaging a rank of 10.7 ^[7].

Clinical Usage

Despite its usage in clinician training, similar to other clinical decision support systems, DXplain has not expanded beyond the research laboratory or medical training setting, due in part to a lack of support by clinicians in real-world settings ^[8].

References