A deep learning algorithm to detect cutaneous squamous cell carcinoma on frozen sections in Mohs micrographic surgery: A retrospective assessment.

Intraoperative margin analysis is crucial for the successful removal of cutaneous squamous cell carcinomas (cSCC). Artificial intelligence technologies (AI) have previously demonstrated potential for facilitating rapid and complete tumour removal using intraoperative margin assessment for basal cell carcinoma. However, the varied morphologies of cSCC present challenges for AI margin assessment. The aim of this study was to develop and evaluate the accuracy of an AI algorithm for real-time histologic margin analysis of cSCC. To do this, a retrospective cohort study was conducted using frozen cSCC section slides. These slides were scanned and annotated, delineating benign tissue structures, inflammation and tumour to develop an AI algorithm for real-time margin analysis. A convolutional neural network workflow was used to extract histomorphological features predictive of cSCC. This algorithm demonstrated proof of concept for identifying cSCC with high accuracy, highlighting the potential for integration of AI into the surgical workflow. Incorporation of AI algorithms may improve efficiency and completeness of real-time margin assessment for cSCC removal, particularly in cases of moderately and poorly differentiated tumours/neoplasms. Further algorithmic improvement incorporating surrounding tissue context is necessary to remain sensitive to the unique epidermal landscape of well-differentiated tumours, and to map tumours to their original anatomical position/orientation.

Potential to Enhance Large Scale Molecular Assessments of Skin Photoaging through Virtual Inference of Spatial Transcriptomics from Routine Staining.

The advent of spatial transcriptomics technologies has heralded a renaissance in research to advance our understanding of the spatial cellular and transcriptional heterogeneity within tissues. Spatial transcriptomics allows investigation of the interplay between cells, molecular pathways, and the surrounding tissue architecture and can help elucidate developmental trajectories, disease pathogenesis, and various niches in the tumor microenvironment. Photoaging is the histological and molecular skin damage resulting from chronic/acute sun exposure and is a major risk factor for skin cancer. Spatial transcriptomics technologies hold promise for improving the reliability of evaluating photoaging and developing new therapeutics. Challenges to current methods include limited focus on dermal elastosis variations and reliance on self-reported measures, which can introduce subjectivity and inconsistency. Spatial transcriptomics offers an opportunity to assess photoaging objectively and reproducibly in studies of carcinogenesis and discern the effectiveness of therapies that intervene in photoaging and preventing cancer. Evaluation of distinct histological architectures using highly-multiplexed spatial technologies can identify specific cell lineages that have been understudied due to their location beyond the depth of UV penetration. However, the cost and interpatient variability using state-of-the-art assays such as the 10x Genomics Spatial Transcriptomics assays limits the scope and scale of large-scale molecular epidemiologic studies. Here, we investigate the inference of spatial transcriptomics information from routine hematoxylin and eosin-stained (H&E) tissue slides. We employed the Visium CytAssist spatial transcriptomics assay to analyze over 18,000 genes at a 50-micron resolution for four patients from a cohort of 261 skin specimens collected adjacent to surgical resection sites for basal cell and squamous cell keratinocyte tumors. The spatial transcriptomics data was co-registered with 40x resolution whole slide imaging (WSI) information. We developed machine learning models that achieved a macro-averaged median AUC and F1 score of 0.80 and 0.61 and Spearman coefficient of 0.60 in inferring transcriptomic profiles across the slides, and accurately captured biological pathways across various tissue architectures.

Development of an interactive web dashboard to facilitate the reexamination of pathology reports for instances of underbilling of CPT codes.

Current Procedural Terminology Codes is a numerical coding system used to bill for medical procedures and services and crucially, represents a major reimbursement pathway. Given that pathology services represent a consequential source of hospital revenue, understanding instances where codes may have been misassigned or underbilled is critical. Several algorithms have been proposed that can identify improperly billed CPT codes in existing datasets of pathology reports. Estimation of the fiscal impacts of these reports requires a coder (i.e., billing staff) to review the original reports and manually code them again. As the re-assignment of codes using machine learning algorithms can be done quickly, the bottleneck in validating these reassignments is in this manual re-coding process, which can prove cumbersome. This work documents the development of a rapidly deployable dashboard for examination of reports that the original coder may have misbilled. Our dashboard features the following main components: (1) a bar plot to show the predicted probabilities for each CPT code, (2) an interpretation plot showing how each word in the report combines to form the overall prediction, and (3) a place for the user to input the CPT code they have chosen to assign. This dashboard utilizes the algorithms developed to accurately identify CPT codes to highlight the codes missed by the original coders. In order to demonstrate the function of this web application, we recruited pathologists to utilize it to highlight reports that had codes incorrectly assigned. We expect this application to accelerate the validation of re-assigned codes through facilitating rapid review of false-positive pathology reports. In the future, we will use this technology to review thousands of past cases in order to estimate the impact of underbilling has on departmental revenue.

Machine learning-based prediction of low-value care for hospitalized patients.

Objective: Low-value care (i.e., costly health care treatments that provide little or no benefit) is an ongoing problem in United States hospitals. Traditional strategies for reducing low-value care are only moderately successful. Informed by behavioral science principles, we sought to use machine learning to inform a targeted prompting system that suggests preferred alternative treatments at the point of care but before clinicians have made a decision. Methods: We used intravenous administration of albumin for fluid resuscitation in intensive care unit (ICU) patients as an exemplar of low-value care practice, identified using the electronic health record of a multi-hospital health system. We divided all ICU episodes into 4-h periods and defined a set of relevant clinical features at the period level. We then developed two machine learning models: a single-stage model that directly predicts if a patient will receive albumin in the next period; and a two-stage model that first predicts if any resuscitation fluid will be administered and then predicts albumin only among the patients with a high probability of fluid use. Results: We examined 87,489 ICU episodes divided into approximately 1.5 million 4-h periods. The area under the receiver operating characteristic curve was 0.86 for both prediction models. The positive predictive value was 0.21 (95% confidence interval: 0.20, 0.23) for the single-stage model and 0.22 (0.20, 0.23) for the two-stage model. Applying either model in a targeted prompting system could prevent 10% of albumin administrations, with an attending physician receiving one prompt every 4.2 days of ICU service. Conclusion: Prediction of low-value care is feasible and could enable a point-of-care, targeted prompting system that offers suggestions ahead of the moment of need before clinicians have already decided. A two-stage approach does not improve performance but does interject new levers for the calibration of such a system.

Potential to Enhance Large Scale Molecular Assessments of Skin Photoaging through Virtual Inference of Spatial Transcriptomics from Routine Staining.

The advent of spatial transcriptomics technologies has heralded a renaissance in research to advance our understanding of the spatial cellular and transcriptional heterogeneity within tissues. Spatial transcriptomics allows investigation of the interplay between cells, molecular pathways and the surrounding tissue architecture and can help elucidate developmental trajectories, disease pathogenesis, and various niches in the tumor microenvironment. Photoaging is the histological and molecular skin damage resulting from chronic/acute sun exposure and is a major risk factor for skin cancer. Spatial transcriptomics technologies hold promise for improving the reliability of evaluating photoaging and developing new therapeutics. Current challenges, including limited focus on dermal elastosis variations and reliance on self-reported measures, can introduce subjectivity and inconsistency. Spatial transcriptomics offer an opportunity to assess photoaging objectively and reproducibly in studies of carcinogenesis and discern the effectiveness of therapies that intervene on photoaging and prevent cancer. Evaluation of distinct histological architectures using highly-multiplexed spatial technologies can identify specific cell lineages that have been understudied due to their location beyond the depth of UV penetration. However, the cost and inter-patient variability using state-of-the-art assays such as the 10x Genomics Spatial Transcriptomics assays limits the scope and scale of large-scale molecular epidemiologic studies. Here, we investigate the inference of spatial transcriptomics information from routine hematoxylin and eosin-stained (H&E) tissue slides. We employed the Visium CytAssist spatial transcriptomics assay to analyze over 18,000 genes at a 50-micron resolution for four patients from a cohort of 261 skin specimens collected adjacent to surgical resection sites for basal and squamous keratinocyte tumors. The spatial transcriptomics data was co-registered with 40x resolution whole slide imaging (WSI) information. We developed machine learning models that achieved a macro-averaged median AUC and F1 score of 0.80 and 0.61 and Spearman coefficient of 0.60 in inferring transcriptomic profiles across the slides, and accurately captured biological pathways across various tissue architectures.

A deep learning algorithm to detect cutaneous squamous cell carcinoma on frozen sections in Mohs micrographic surgery: a retrospective assessment.

Importance: Intraoperative margin analysis is crucial for the successful removal of cutaneous squamous cell carcinomas (cSCC). Artificial intelligence technologies (AI) have previously demonstrated potential for facilitating rapid and complete tumor removal using intraoperative margin assessment for basal cell carcinoma. However, the varied morphologies of cSCC present challenges for AI margin assessment. Objective: To develop and evaluate the accuracy of an AI algorithm for real-time histologic margin analysis of cSCC. Design: A retrospective cohort study was conducted using frozen cSCC section slides and adjacent tissues. Setting: This study was conducted in a tertiary care academic center. Participants: Patients undergoing Mohs micrographic surgery for cSCC between January and March 2020. Exposures: Frozen section slides were scanned and annotated, delineating benign tissue structures, inflammation, and tumor to develop an AI algorithm for real-time margin analysis. Patients were stratified by tumor differentiation status. Epithelial tissues including epidermis and hair follicles were annotated for moderate-well to well differentiated cSCC tumors. A convolutional neural network workflow was used to extract histomorphological features predictive of cSCC at 50-micron resolution. Main Outcomes and Measures: The performance of the AI algorithm in identifying cSCC at 50-micron resolution was reported using the area under the receiver operating characteristic curve. Accuracy was also reported by tumor differentiation status and by delineation of cSCC from epidermis. Model performance using histomorphological features alone was compared to architectural features (i.e., tissue context) for well-differentiated tumors. Results: The AI algorithm demonstrated proof of concept for identifying cSCC with high accuracy. Accuracy differed by differentiation status, driven by challenges in separating cSCC from epidermis using histomorphological features alone for well-differentiated tumors. Consideration of broader tissue context through architectural features improved the ability to delineate tumor from epidermis. Conclusions and Relevance: Incorporating AI into the surgical workflow may improve efficiency and completeness of real-time margin assessment for cSCC removal, particularly in cases of moderately and poorly differentiated tumors/neoplasms. Further algorithmic improvement is necessary to remain sensitive to the unique epidermal landscape of well-differentiated tumors, and to map tumors to their original anatomical position/orientation. Future studies should assess the efficiency improvements and cost benefits and address other confounding pathologies such as inflammation and nuclei. Funding sources: JL is supported by NIH grants R24GM141194, P20GM104416 and P20GM130454. Support for this work was also provided by the Prouty Dartmouth Cancer Center development funds. Key Points: Question: How can the efficiency and accuracy of real-time intraoperative margin analysis for the removal of cutaneous squamous cell carcinoma (cSCC) be improved, and how can tumor differentiation be incorporated into this approach?Findings: A proof-of-concept deep learning algorithm was trained, validated, and tested on frozen section whole slide images (WSI) for a retrospective cohort of cSCC cases, demonstrating high accuracy in identifying cSCC and related pathologies. Histomorphology alone was found to be insufficient to delineate tumor from epidermis in histologic identification of well-differentiated cSCC. Incorporation of surrounding tissue architecture and shape improved the ability to delineate tumor from normal tissue.Meaning: Integrating artificial intelligence into surgical procedures has the potential to enhance the thoroughness and efficiency of intraoperative margin analysis for cSCC removal. However, accurately accounting for the epidermal tissue based on the tumor's differentiation status requires specialized algorithms that consider the surrounding tissue context. To meaningfully integrate AI algorithms into clinical practice, further algorithmic refinement is needed, as well as the mapping of tumors to their original surgical site, and evaluation of the cost and efficacy of these approaches to address existing bottlenecks.

Identifying the Perceived Severity of Patient-Generated Telemedical Queries Regarding COVID: Developing and Evaluating a Transfer Learning-Based Solution.

BACKGROUND: Triage of textual telemedical queries is a safety-critical task for medical service providers with limited remote health resources. The prioritization of patient queries containing medically severe text is necessary to optimize resource usage and provide care to those with time-sensitive needs. OBJECTIVE: We aim to evaluate the effectiveness of transfer learning solutions on the task of telemedical triage and provide a thorough error analysis, identifying telemedical queries that challenge state-of-the-art natural language processing (NLP) systems. Additionally, we aim to provide a publicly available telemedical query data set with labels for severity classification for telemedical triage of respiratory issues. METHODS: We annotated 573 medical queries from 3 online health platforms: HealthTap, HealthcareMagic, and iCliniq. We then evaluated 6 transfer learning solutions utilizing various text-embedding strategies. Specifically, we first established a baseline using a lexical classification model with term frequency-inverse document frequency (TF-IDF) features. Next, we investigated the effectiveness of global vectors for text representation (GloVe), a pretrained word-embedding method. We evaluated the performance of GloVe embeddings in the context of support vector machines (SVMs), bidirectional long short-term memory (bi-LSTM) networks, and hierarchical attention networks (HANs). Finally, we evaluated the performance of contextual text embeddings using transformer-based architectures. Specifically, we evaluated bidirectional encoder representation from transformers (BERT), Bio+Clinical-BERT, and Sentence-BERT (SBERT) on the telemedical triage task. RESULTS: We found that a simple lexical model achieved a mean F1 score of 0.865 (SD 0.048) on the telemedical triage task. GloVe-based models using SVMs, HANs, and bi-LSTMs achieved a 0.8-, 1.5-, and 2.1-point increase in the F1 score, respectively. Transformer-based models, such as BERT, Bio+Clinical-BERT, and SBERT, achieved a mean F1 score of 0.914 (SD 0.034), 0.904 (SD 0.041), and 0.917 (SD 0.037), respectively. The highest-performing model, SBERT, provided a statistically significant improvement compared to all GloVe-based and lexical baselines. However, no statistical significance was found when comparing transformer-based models. Furthermore, our error analysis revealed highly challenging query types, including those with complex negations, temporal relationships, and patient intents. CONCLUSIONS: We showed that state-of-the-art transfer learning techniques work well on the telemedical triage task, providing significant performance increase over lexical models. Additionally, we released a public telemedical triage data set using labeled questions from online medical question-and-answer (Q&A) platforms. Our analysis highlights various avenues for future works that explicitly model such query challenges.

Validity and Feasibility of the Monitoring and Modeling Family Eating Dynamics System to Automatically Detect In-field Family Eating Behavior: Observational Study.

BACKGROUND: The field of dietary assessment has a long history, marked by both controversies and advances. Emerging technologies may be a potential solution to address the limitations of self-report dietary assessment methods. The Monitoring and Modeling Family Eating Dynamics (M2FED) study uses wrist-worn smartwatches to automatically detect real-time eating activity in the field. The ecological momentary assessment (EMA) methodology was also used to confirm whether eating occurred (ie, ground truth) and to measure other contextual information, including positive and negative affect, hunger, satiety, mindful eating, and social context. OBJECTIVE: This study aims to report on participant compliance (feasibility) to the 2 distinct EMA protocols of the M2FED study (hourly time-triggered and eating event-triggered assessments) and on the performance (validity) of the smartwatch algorithm in automatically detecting eating events in a family-based study. METHODS: In all, 20 families (58 participants) participated in the 2-week, observational, M2FED study. All participants wore a smartwatch on their dominant hand and responded to time-triggered and eating event-triggered mobile questionnaires via EMA while at home. Compliance to EMA was calculated overall, for hourly time-triggered mobile questionnaires, and for eating event-triggered mobile questionnaires. The predictors of compliance were determined using a logistic regression model. The number of true and false positive eating events was calculated, as well as the precision of the smartwatch algorithm. The Mann-Whitney U test, Kruskal-Wallis test, and Spearman rank correlation were used to determine whether there were differences in the detection of eating events by participant age, gender, family role, and height. RESULTS: The overall compliance rate across the 20 deployments was 89.26% (3723/4171) for all EMAs, 89.7% (3328/3710) for time-triggered EMAs, and 85.7% (395/461) for eating event-triggered EMAs. Time of day (afternoon odds ratio [OR] 0.60, 95% CI 0.42-0.85; evening OR 0.53, 95% CI 0.38-0.74) and whether other family members had also answered an EMA (OR 2.07, 95% CI 1.66-2.58) were significant predictors of compliance to time-triggered EMAs. Weekend status (OR 2.40, 95% CI 1.25-4.91) and deployment day (OR 0.92, 95% CI 0.86-0.97) were significant predictors of compliance to eating event-triggered EMAs. Participants confirmed that 76.5% (302/395) of the detected events were true eating events (ie, true positives), and the precision was 0.77. The proportion of correctly detected eating events did not significantly differ by participant age, gender, family role, or height (P>.05). CONCLUSIONS: This study demonstrates that EMA is a feasible tool to collect ground-truth eating activity and thus evaluate the performance of wearable sensors in the field. The combination of a wrist-worn smartwatch to automatically detect eating and a mobile device to capture ground-truth eating activity offers key advantages for the user and makes mobile health technologies more accessible to nonengineering behavioral researchers.