Case Presentation: A female of 70-years with a history of osteoarthritis presented with four weeks of fever, weight loss, night sweats, and fatigue. She was hypotensive, febrile, tachycardic and tachypneic. She was initially managed per sepsis protocol. Laboratory data revealed WBC 2.7 K/UL, Hgb 9.8 g/dL, PLT 77 K/DL, AST 71 IU/L ALT 49 IU/L, CRP 12 mg/dL, ESR 41 mm/hr. Ferritin 7138 ng/mL, LDH 570 IU/L. She continued to be febrile, pancytopenic and generally ill for the first 48 hours despite IV antibiotics. A contrast-enhanced CT of the abdomen revealed splenic hypodensities; laboratory workup for Q-fever, RMSF, salmonella, and Lyme’s disease was sent to evaluate for secondary hemophagocytic lymphohistiocytosis (HLH).Lack of clinical and laboratory improvement prompted us to utilize Isabel Healthcare Differential Diagnosis (DDX) Tool, a resource available in our institution. Our keywords generated ten differential diagnosis including Brucellosis. Our patient later admitted consuming homemade cheese on her last visit to Mexico. Brucella serology, blood cultures, and a liver biopsy were obtained. Her fever resolved 48 hours after doxycycline was initiated. Blood cultures grew Brucella spp. Currently, our patient remains symptom-free and all laboratory abnormalities have normalized.

Discussion: Medical decision making relies on pattern recognition of illness scripts accumulated over time. Illness scripts can be contaminated by our own heuristics and biases. Complex and rare conditions like HLH secondary to Brucellosis may require more analytical decision making with the establishment of true probability. Historically, we referred to textbooks for the DDX for each symptom. Nowadays, we have more sophisticated DDX generators that include epidemiological data, clinical and laboratory findings to generate a DDX list. In a study, the pooled accurate diagnosis retrieval rate for DDX tools was as high as 0.70. To be able to utilize a DDX generator properly, it is crucial to identify significant and clinically relevant data. The expert physician goes through a series of hypothesis refinement during the process of solving a complex case and the search might be more refined to find the most probable etiologies. More studies are needed to test these systems’ ability to provide a final diagnosis. At this time, DDX generators may be useful resources for clinicians and it may also increase the medical knowledge.

Conclusions: Multiple technologic efforts have been made to objectively synthesize and interpret the data originating from medical records. Clinical decision support tools are intended to provide health care targeted and timely information to improve clinical decisions to optimize healthcare delivery. We can conclude that further research is needed in this regard to decrease excessive reliance on these tools as well as to identify and minimize flaws such as bias that may lead to socioeconomic disparities in health care.