AI-clinician collaboration via disagreement prediction: A decision pipeline and retrospective analysis of real-world radiologist-AI interactions.

Clinical decision support tools can improve diagnostic performance or reduce variability, but they are also subject to post-deployment underperformance. Although using AI in an assistive setting offsets many concerns with autonomous AI in medicine, systems that present all predictions equivalently fail to protect against key AI safety concerns. We design a decision pipeline that supports the diagnostic model with an ecosystem of models, integrating disagreement prediction, clinical significance categorization, and prediction quality modeling to guide prediction presentation. We characterize disagreement using data from a deployed chest X-ray interpretation aid and compare clinician burden in this proposed pipeline to the diagnostic model in isolation. The average disagreement rate is 6.5%, and the expected burden reduction is 4.8%, even if 5% of disagreements on urgent findings receive a second read. We conclude that, in our production setting, we can adequately balance risk mitigation with clinician burden if disagreement false positives are reduced.

Wavelet radiomics features from multiphase CT images for screening hepatocellular carcinoma: analysis and comparison.

Early detection of liver malignancy based on medical image analysis plays a crucial role in patient prognosis and personalized treatment. This task, however, is challenging due to several factors, including medical data scarcity and limited training samples. This paper presents a study of three important aspects of radiomics feature from multiphase computed tomography (CT) for classifying hepatocellular carcinoma (HCC) and other focal liver lesions: wavelet-transformed feature extraction, relevant feature selection, and radiomics features-based classification under the inadequate training samples. Our analysis shows that combining radiomics features extracted from the wavelet and original CT domains enhance the classification performance significantly, compared with using those extracted from the wavelet or original domain only. To facilitate the multi-domain and multiphase radiomics feature combination, we introduce a logistic sparsity-based model for feature selection with Bayesian optimization and find that the proposed model yields more discriminative and relevant features than several existing methods, including filter-based, wrapper-based, or other model-based techniques. In addition, we present analysis and performance comparison with several recent deep convolutional neural network (CNN)-based feature models proposed for hepatic lesion diagnosis. The results show that under the inadequate data scenario, the proposed wavelet radiomics feature model produces comparable, if not higher, performance metrics than the CNN-based feature models in terms of area under the curve.

Yes, we CAM: Increasing physician communication about complementary and alternative medicine through electronic medical record prompts in a pediatric hospital.

BACKGROUND AND PURPOSE: Over half of patients utilize complementary and alternative medicine (CAM), yet fewer than 10 % of physicians inquire about use. This prospective cohort study sought to increase patient-provider communication about CAM through electronic medical record (EMR) prompts and education, as well as study concordance rates of physician and nursing CAM-related documentation. MATERIALS AND METHODS: Recordings in history and physical (H&P) documents authored by first-year pediatric residents were used as a proxy for communication. Rates of documentation were assessed at baseline, after the introduction of an EMR prompt, and after an educational intervention. Nursing documentation was compared with resident documents to assess rates of concordance regarding CAM-related documentation. RESULTS: Baseline CAM-related documentation rate was 24 % and increased to 50 % after introducing an EMR CAM prompt (p < 0.001). No significant change occurred after education: 38 % had CAM-related documentation (p = 0.09). Physician and nursing documentation concordance rates were 58 % at baseline, 48 % after introduction of prompts, and 35 % after introducing education. CONCLUSION: Visual cues alone may be effective in increasing patient-provider communication about CAM, though low concordance between physician and nursing documentation may suggest variability in how CAM is defined and inquired about.

Symptomatic perioperative venous thromboembolism is a frequent complication in patients with a history of deep vein thrombosis.

OBJECTIVES: Patients who undergo surgery are at risk for venous thromboembolism (VTE), and a history of prior deep vein thrombosis (DVT) increases that risk. This study determined the incidence and risk factors for symptomatic perioperative VTE in patients with a prior diagnosis of DVT. METHODS: All lower extremity DVTs, diagnosed between January 2002 and December 2006, were identified through a vascular database. Patients who had subsequent surgery were reviewed. The following data were evaluated: location of DVT, time interval between DVT and surgery, type of surgery, common clinical VTE risk factors, postoperative venous duplex scans, computed tomography (CT) scans of the chest, and ventilation-perfusion scans. RESULTS: A total of 372 patients with prior DVT underwent 1081 subsequent surgical procedures. One hundred nine patients undergoing 211 procedures had a follow-up venous duplex scan within 30 days after surgery. Of them, 46% received an inferior vena caval (IVC) filter, and pulmonary emboli were diagnosed in 3 patients (<1%). Overall, 24% of the patients developed DVT extension or new-site DVT in the perioperative period. The median time interval between the original DVT and surgery was 1.5 weeks in patients with DVT recurrence and 4 weeks in patients without recurrence (P = .22, Mann-Whitney). High-risk surgeries were associated with a >three-fold increased risk for recurrence, when compared with low-risk procedures (34% vs 11%; P = .009, chi(2)). Perioperative VTE recurrence was not influenced by the location of the original thrombus or other VTE risk factors. CONCLUSION: In patients with prior DVT, perioperative symptomatic recurrence is common and is associated with high-risk procedures. A longer time interval between a DVT episode and subsequent surgery may decrease the risk of recurrence, but large clinical trials are needed to confirm this. Further prospective evaluations are needed to identify and treat patients at greatest risk for recurrence.