Riverview Medical Center

DeepSeek is a newly introduced large language model (LLM) designed for enhanced reasoning, but its medical-domain capabilities have not yet been evaluated. Here we assessed the capabilities of three LLMs- DeepSeek-R1, ChatGPT-o1 and Llama 3.1-405B-in performing four different medical tasks: answering questions from the United States Medical Licensing Examination (USMLE), interpreting and reasoning on the basis of text-based diagnostic and management cases, providing tumor classification according to RECIST 1.1 criteria and providing summaries of diagnostic imaging reports across multiple modalities. In the USMLE test, the performance of DeepSeek-R1 (accuracy 0.92) was slightly inferior to that of ChatGPT-o1 (accuracy 0.95; P = 0.04) but better than that of Llama 3.1-405B (accuracy 0.83; P < 10-3). For text-based case challenges, DeepSeek-R1 performed similarly to ChatGPT-o1 (accuracy of 0.57 versus 0.55; P = 0.76 and 0.74 versus 0.76; P = 0.06, using New England Journal of Medicine and Médicilline databases, respectively). For RECIST classifications, DeepSeek-R1 also performed similarly to ChatGPT-o1 (0.74 versus 0.81; P = 0.10). Diagnostic reasoning steps provided by DeepSeek were deemed more accurate than those provided by ChatGPT and Llama 3.1-405B (average Likert score of 3.61, 3.22 and 3.13, respectively, P = 0.005 and P < 10-3). However, summarized imaging reports provided by DeepSeek-R1 exhibited lower global quality than those provided by ChatGPT-o1 (5-point Likert score: 4.5 versus 4.8; P < 10-3). This study highlights the potential of DeepSeek-R1 LLM for medical applications but also underlines areas needing improvements.

Artificial intelligence (AI), particularly deep learning, has made enormous strides in medical imaging analysis. In the field of musculoskeletal radiology, deep-learning models are actively being developed for the identification and evaluation of bone fractures. These methods provide numerous benefits to radiologists such as increased diagnostic accuracy and efficiency while also achieving standalone performances comparable or superior to clinician readers. Various algorithms are already commercially available for integration into clinical workflows, with the potential to improve healthcare delivery and shape the future practice of radiology. In this systematic review, we explore the performance of current AI methods in the identification and evaluation of fractures, particularly those in the ankle, wrist, hip, and ribs. We also discuss current commercially available products for fracture detection and provide an overview of the current limitations of this technology and future directions of the field.

BackgroundAutomatic noninvasive oscillometric blood pressure (NIBP) devices measure mean arterial pressure (MAP); systolic and diastolic blood pressure (SBP, DBP) are algorithmically derived from MAP. The most invalid NIBP measure is SBP, yet stroke practitioners use it to manage blood pressure (BP) in accordance with thrombolysis guidelines. We determined agreement between SBP, DBP, and MAP measured manually and by NIBP in patients treated with alteplase.MethodsA multisite prospective observational study of NIBP and manual BP agreement was conducted in patients treated with alteplase immediately after bolus and infusion initiation using methods established in guidelines for the assessment of device agreement. Dual auscultatory stethoscopes were used by 2 investigators to ensure agreement with each manual BP variable and MAP was calculated using the standard formula for manual BP measures. Data were analyzed using Bland–Altman analyses and Lin concordance correlation coefficient.ResultsA total of 7 hospitals participated, collecting 5 sets of manual/NIBP BPs in 95 patients treated with alteplase (475 paired measures). Range in limits of agreement were SBP: −28.91 to 21.41 mmHg with Lin's concordance correlation coefficient 0.8; DBP: −21.0 to 19.0 mmHg with Lin's concordance correlation coefficient 0.69; and MAP: −27.5 to 16.5 mmHg with Lin's concordance correlation coefficient 0.7. There was no difference in device agreement by BP device manufacturer brand. Differences in SBP, DBP, and MAP between NIBP and manual sphygmomanometry failed to reach guideline recommendations requiring 80% of measures to fall within a 5 mmHg difference and 95% of measures to fall within a 10 mmHg difference.ConclusionNIBP devices produce significantly different BP measures then manual sphygmomanometry auscultated BP. Because NIBP devices rely on the MAP and do not directly measure SBP and DBP, definition of what constitutes safe MAP boundaries in patients treated with alteplase should be determined when automatic BP measurement is used in clinical practice.

We present an atypical case of secondary syphilis concomitant with condyloma lata in an HIV-positive male. The patient presented with lesions on his lateral neck, soles, and palms. Spirochetes were examined on the biopsy of the palm, consistent with secondary syphilis, but not on the biopsy from the neck. The lesion on the neck was diagnosed as an evolving secondary syphilis lesion due to the extent of the lymphoplasmacytic infiltration, presence of acanthosis and epidermal hyperplasia, the accompanying diagnosis of secondary syphilis on the palm, the patient’s history of HIV, and the related diagnosis of condyloma lata on the dorsal shaft of the penis. Following this diagnosis, the patient was referred to an infectious disease specialist who treated him with a single intramuscular injection of 2.4 million units of benzathine penicillin, per CDC guidelines. This case highlights the importance of clinicians taking into consideration a variety of factors when diagnosing and treating cutaneous disorders with a broad-spectrum differential diagnosis, particularly in at-risk individuals.

Severe traumatic injury (sTBI) continues to be a common source of morbidity and mortality. While there have been several advances in understanding the pathophysiology of this injury, the clinical outcome has remained grim. These trauma patients often require multidisciplinary care and are admitted to a surgical service line, depending on hospital policy. A retrospective chart review spanning 2019-2022 was completed using the electronic health record of the neurosurgery service. We identified 140 patients with a Glasgow Coma Scale (GCS) of eight or less, ages 18-99, who were admitted to a level-one trauma center in Southern California. Seventy patients were admitted under the neurosurgery service, while the other half were admitted to the surgical intensive care unit (SICU) service after initial assessment in the emergency department by both services to evaluate for multisystem injury. Between both groups, the injury severity scores that evaluated patients' overall injuries were not significantly different. The results demonstrate a significant difference in GCS change, modified Rankin Scale (mRS) change, and Glasgow Outcome Scale (GOS) change between the two groups. Furthermore, the mortality rate differed between neurosurgical care and other service care by 27% and 51%, respectively, despite similar Injury Severity Scores (ISS) (p=0.0026). Therefore, this data demonstrates that a well-trained neurosurgeon with critical care experience can safely manage a severe traumatic brain injury patient with an isolated head injury as a primary service while in the intensive care unit. Since injury severity scores did not differ between these two service lines, we further theorize that this is likely due to a deep understanding of the nuances of neurosurgical pathophysiology and Brain Trauma Foundation (BTF) guidelines.

Critical Care Medicine: January 2020 - Volume 48 - Issue 1 - p 468 doi: 10.1097/01.ccm.0000633444.89554.8d