The necessity for explainability of artificial intelligence technologies in medical applications has been widely discussed and heavily debated within the literature. This paper comprises a systematized review of the arguments supporting and opposing this purported necessity. Both sides of the debate within the literature are quoted to synthesize discourse on common recurring themes and subsequently critically analyze and respond to it. While the use of autonomous black box algorithms is compellingly discouraged, the same cannot be said for the whole of medical artificial intelligence technologies that lack explainability. We contribute novel comparisons of unexplainable clinical artificial intelligence tools, diagnosis of idiopathy, and diagnoses by exclusion, to analyze implications on patient autonomy and informed consent. Applying a novel approach using comparisons with clinical practice guidelines, we contest the claim that lack of explainability compromises clinician due diligence and undermines epistemological responsibility. We find it problematic that many arguments in favour of the practical, ethical, or legal necessity of clinical artificial intelligence explainability conflate the use of unexplainable AI with automated decision making, or equate the use of clinical artificial intelligence with the exclusive use of clinical artificial intelligence.
3 June 2025 in Journal of Orthopaedic Surgery and Research
Background: Statistical Shape Models are machine learning tools in computational orthopedics that enable the study of anatomical variability and the creation of synthetic models for pathogenetic analysis and surgical planning. Current models of the glenohumeral joint either describe individual bones or are limited to non-pathologic datasets, failing to capture coupled shape variation in arthropathic anatomy. We aimed to develop a novel combined scapula-proximal-humerus model applicable to clinical populations. Methods: Preoperative computed tomography scans from 45 Reverse Total Shoulder Arthroplasty patients were used to generate three-dimensional models of the scapula and proximal humerus. Correspondence point clouds were combined into a two-body shape model using Principal Component Analysis. Individual scapula-only and proximal-humerus-only shape models were also created for comparison. The models were validated using compactness, specificity, generalization ability, and leave-one-out cross-validation. The modes of variation for each model were also compared. The combined model was described using eigenvector decomposition into single body models. The models were further compared in their ability to predict the shape of one body when given the shape of its counterpart, and the generation of diverse realistic synthetic pairs de novo. Results: The scapula and proximal-humerus models performed comparably to previous studies with median average leave-one-out cross-validation errors of 1.08 mm (IQR: 0.359 mm), and 0.521 mm (IQR: 0.111 mm); the combined model was similar with median error of 1.13 mm (IQR: 0.239 mm). The combined model described coupled variations between the shapes equalling 43.2% of their individual variabilities, including the relationship between glenoid and humeral head erosions. The combined model outperformed the individual models generatively with reduced missing shape prediction bias (> 10%) and uniformly diverse shape plausibility (uniformity p-value < .001 vs. .59). Conclusions: This study developed the first two-body scapulohumeral shape model that captures coupled variations in arthropathic shoulder anatomy and the first proximal-humeral statistical model constructed using a clinical dataset. While single-body models are effective for descriptive tasks, combined models excel in generating joint-level anatomy. This model can be used to augment computational analyses of synthetic populations investigating shoulder biomechanics and surgical planning.
Central nervous system involvement in multiple myeloma (MM‐CNS) is a condition with poor prognosis and no clear treatment options. Standard regimens, including proteasome inhibitors (PIs) and immunomodulatory (IMiD) agents, provide minimal benefit in this setting, highlighting the need for novel therapies. Teclistamab, a bispecific T‐cell engager (BiTE) targeting B‐cell maturation agent (BCMA) and CD3, has demonstrated robust systemic activity in heavily pretreated MM but its role in CNS disease remains undefined, as patients with CNS involvement have been excluded from pivotal trials. We present the case of a 62‐year‐old female with high‐risk MM who developed extensive leptomeningeal myelomatosis following multiple lines of therapy including autologous transplantation, PI‐ and IMiD‐based regimens, and palliative radiotherapy. Upon presentation with confusion, aphasia, and ataxia, MRI revealed diffuse leptomeningeal enhancement. The patient elected to proceed with teclistamab therapy. Following two cycles, she achieved a very good partial serologic response and MRI demonstrated marked radiologic improvement with resolution of cerebellar nodularity and sulcal enhancement. However, functional recovery was not observed, and the treatment was discontinued after three cycles due to clinical decline and infectious complications. She subsequently transitioned to supportive care and passed away 1 month later. This case report documents one of the first reports of teclistamab demonstrating radiologic improvement in leptomeningeal disease in MM‐CNS. While the patient’s overall outcome was poor, the observed CNS response supports the biologic plausibility of BiTE penetration and activity in the CNS. These findings suggest the urgent need for prospective studies of BCMA‐directed bispecific antibodies in MM‐CNS, as well as earlier intervention prior to functional decline.
22 July 2025 in Journal of Vascular Surgery Cases, Innovations and Techniques
Postimplantation syndrome (PIS) is a noninfectious systemic inflammatory response that often follows endovascular aneurysm repair, historically presumed to be a reaction to implanted graft materials. Although well-documented with stent graft use, PIS in the absence of stent grafts remains poorly understood. We present an instance of PIS after stentless thoracic endovascular aortic repair, which involves embolizing intimal tears with plugs and/or coils. This case highlights the need for further investigation into the pathophysiology of PIS. PIS can occur in the absence of a stent graft; prophylactic anti-inflammatory therapy may be warranted during endovascular interventions with large-volume aneurysms or false lumen thrombosis.
Background Percutaneous coronary intervention (PCI) for acute coronary syndrome (ACS) is central to modern cardiac care, yet the factors determining how many vessels are treated during a procedure (‘PCI Extensiveness’) remain poorly understood, particularly in fee-for-service systems that reward more extensive treatment. This study aimed to evaluate the association between various factors and PCI Extensiveness. Methods We conducted a multicentre retrospective study using provincial fee-for-service billing data, which pays incremental fees per vessel treated from all adult cardiac catheterisation laboratories in British Columbia, Canada (population 5.5 million). We included 72 390 cath lab encounters for presumed ST-Elevation Myocardial Infarction (STEMI) or Non-ST Elevation ACS (NSTE-ACS) between 2013 and 2020 who underwent angiogram ± PCI. Patient-specific factors (age, comorbidities, prior myocardial infarction, prior coronary interventions, ACS type and angiographic findings) were incorporated into a non-parametric machine-learning model to estimate the number of vessels treated based on case mix. Results Interventionalists varied widely in mean vessels treated per case (0.455–3.19), with remarkable internal consistency per operator across STEMI and NSTE-ACS cases (R=0.956; 95% CI 0.904 to 0.981; p<0.001). However, they had similar angiographic case mixes (Cramér’s V≤0.0955). The machine-learning patient factor model suggested that interventionalists should have only ranged from 0.976 to 1.42 based on differences in case mix, 6.10-fold narrower (range across imputations: 6.03–6.16) than actually observed. Between-hospital differences could only explain 45.6% of the observed variation between interventionalists. Results were consistent when considering only the index cath lab encounter or all PCI performed within 60 days subsequently. Conclusions In contemporary ACS care, who performs the procedure appears to be associated with how extensively a patient is treated. In a system that financially rewards more extensive treatment, this unexplained operator-level variation raises the hypothesis that compensation structures may shape treatment intensity independent of clinical need.
The objective of this study is to evaluate whether ChatGPT models can reliably apply the DISCERN instrument, a 16-question human-scored rubric developed in 1999 to evaluate consumer health information, and assess the impact of prompting strategies, model choice, and scoring repeatability on agreement with human-derived DISCERN scores. A PubMed search of “DISCERN” identified English-language studies since 2019 reporting exact webpage URLs with corresponding human-derived DISCERN scores. Archived versions of 42 webpages were retrieved. Three ChatGPT models (GPT-5.2, GPT-4o, and o3) were evaluated using four prompting strategies: “Naïve” zero-shot, item-level “Split” scoring, “Augmented” prompting with DISCERN guidance, and a “Combined” split-plus-augmented approach. Agreement with human scores was assessed using correlations and absolute differences. Repeatability was examined using 10 repeated scoring runs across 9 webpages. Agreement between ChatGPT-generated and human DISCERN scores was weak to moderate. All models demonstrated systematic score compression, overestimating low-quality webpages and underestimating high-quality webpages. Combined prompting modestly improved agreement and reduced absolute error, particularly for the o3 model, which consistently outperformed GPT-5.2 and GPT-4o. Substantial run-to-run variability was observed with a mean score range of 17.5 points and ranges up to 43 points for the same webpage. Averaging scores across runs did not improve agreement with human ratings. ChatGPT's DISCERN scoring reflects systematic attenuation consistent with prediction under noisy subjective measurement. Prompt engineering did not correct calibration bias or reproducibility limitations. Under the prompting strategies evaluated, ChatGPT models were insufficient for reliable automated DISCERN scoring. Persistent attenuation bias and poor repeatability significantly limit clinical or research applicability.