Matthew Michelson

Background: Rapid access to evidence is crucial in times of an evolving clinical crisis. To that end, we propose a novel approach to answer clinical queries, termed rapid meta-analysis (RMA). Unlike traditional meta-analysis, RMA balances a quick time to production with reasonable data quality assurances, leveraging artificial intelligence (AI) to strike this balance. Objective: We aimed to evaluate whether RMA can generate meaningful clinical insights, but crucially, in a much faster processing time than traditional meta-analysis, using a relevant, real-world example. Methods: The development of our RMA approach was motivated by a currently relevant clinical question: is ocular toxicity and vision compromise a side effect of hydroxychloroquine therapy? At the time of designing this study, hydroxychloroquine was a leading candidate in the treatment of coronavirus disease (COVID-19). We then leveraged AI to pull and screen articles, automatically extract their results, review the studies, and analyze the data with standard statistical methods. Results: By combining AI with human analysis in our RMA, we generated a meaningful, clinical result in less than 30 minutes. The RMA identified 11 studies considering ocular toxicity as a side effect of hydroxychloroquine and estimated the incidence to be 3.4% (95% CI 1.11%-9.96%). The heterogeneity across individual study findings was high, which should be taken into account in interpretation of the result. Conclusions: We demonstrate that a novel approach to meta-analysis using AI can generate meaningful clinical insights in a much shorter time period than traditional meta-analysis.

BACKGROUND Rapid access to evidence is crucial in times of an evolving clinical crisis. To that end, we propose a novel approach to answer clinical queries, termed rapid meta-analysis (RMA). Unlike traditional meta-analysis, RMA balances a quick time to production with reasonable data quality assurances, leveraging artificial intelligence (AI) to strike this balance. OBJECTIVE We aimed to evaluate whether RMA can generate meaningful clinical insights, but crucially, in a much faster processing time than traditional meta-analysis, using a relevant, real-world example. METHODS The development of our RMA approach was motivated by a currently relevant clinical question: is ocular toxicity and vision compromise a side effect of hydroxychloroquine therapy? At the time of designing this study, hydroxychloroquine was a leading candidate in the treatment of coronavirus disease (COVID-19). We then leveraged AI to pull and screen articles, automatically extract their results, review the studies, and analyze the data with standard statistical methods. RESULTS By combining AI with human analysis in our RMA, we generated a meaningful, clinical result in less than 30 minutes. The RMA identified 11 studies considering ocular toxicity as a side effect of hydroxychloroquine and estimated the incidence to be 3.4% (95% CI 1.11%-9.96%). The heterogeneity across individual study findings was high, which should be taken into account in interpretation of the result. CONCLUSIONS We demonstrate that a novel approach to meta-analysis using AI can generate meaningful clinical insights in a much shorter time period than traditional meta-analysis.

Rapid access to evidence is crucial in times of evolving clinical crisis. To that end, we propose a novel mechanism to answer clinical queries: Rapid Meta-Analysis (RMA). Unlike traditional meta-analysis, RMA balances quick time-to-production with reasonable data quality assurances, leveraging Artificial Intelligence to strike this balance. This article presents an example RMA to a currently relevant clinical question: Is ocular toxicity and vision compromise a side effect with hydroxychloroquine therapy? As of this writing, hydroxychloroquine is a leading candidate in the treatment of COVID-19. By combining AI with human analysis, our RMA identified 11 studies looking at ocular toxicity as a side effect and estimated the incidence to be 3.4% (95% CI: 1.11-9.96%). The heterogeneity across the individual study findings was high, and interpretation of the result should take this into account. Importantly, this RMA, from search to screen to analysis, took less than 30 minutes to produce.

Journal of Medical Internet Research - International Scientific Journal for Medical Research, Information and Communication on the Internet #Preprint #PeerReviewMe: Warning: This is a unreviewed preprint. Readers are warned that the document has not been peer-reviewed by expert/patient reviewers or an academic editor, may contain misleading claims, and is likely to undergo changes before final publication, if accepted, or may have been rejected/withdrawn. Readers with interest and expertise are encouraged to sign up as peer-reviewer, if the paper is within an open peer-review period. Please cite this preprint only for review purposes or for grant applications and CVs (if you are the author). Background: A major barrier to the practice of evidence-based medicine is efficiently finding scientifically sound studies on a given clinical topic. Objective: To investigate a deep learning approach to retrieve scientifically sound treatment studies from the biomedical literature. Methods: We trained a Convolutional Neural Network using a noisy dataset of 403,216 PubMed citations with title and abstract as features. The deep learning model was compared with state-of-the-art search filters, i.e. PubMed's Clinical Query Broad treatment filter, McMaster's textword search strategy (no MeSH terms), and Clinical Query Balanced treatment filter. A previously annotated dataset (Clinical Hedges) was used as the gold standard. Results: The deep learning model obtained significantly lower recall than the Clinical Queries Broad treatment filter (96.9% vs. 98.4%; p<0.001); and equivalent to McMaster's textword search (96.9% vs. 97.1%; p=0.57) and Clinical Queries Balanced filter (96.9% vs. 97.0%; p=0.63). Deep learning obtained significantly higher precision than the Clinical Queries Broad filter (34.6% vs. 22.4%; p<0.001) and McMaster's textword search (34.6% vs. 11.8%; p<0.001); but significantly lower than the Clinical Queries Balanced filter (34.6% vs. 40.9%; p=<0.001). Conclusions: Deep learning performed well compared to state-of-the-art search filters, especially when citations are not indexed. Unlike previous machine learning approaches, the proposed deep learning model does not require feature engineering, time-sensitive or proprietary features, such as MeSH terms and bibliometrics. Deep learning is a promising approach to identifying reports of scientifically rigorous clinical research. Further work is needed to optimize the deep learning model and to assess generalizability to other areas, such as diagnosis, etiology, and prognosis.

BACKGROUND A major barrier to the practice of evidence-based medicine is efficiently finding scientifically sound studies on a given clinical topic. OBJECTIVE To investigate a deep learning approach to retrieve scientifically sound treatment studies from the biomedical literature. METHODS We trained a Convolutional Neural Network using a noisy dataset of 403,216 PubMed citations with title and abstract as features. The deep learning model was compared with state-of-the-art search filters, such as PubMed’s Clinical Query Broad treatment filter, McMaster’s textword search strategy (no Medical Subject Heading, MeSH, terms), and Clinical Query Balanced treatment filter. A previously annotated dataset (Clinical Hedges) was used as the gold standard. RESULTS The deep learning model obtained significantly lower recall than the Clinical Queries Broad treatment filter (96.9% vs 98.4%; P<.001); and equivalent recall to McMaster’s textword search (96.9% vs 97.1%; P=.57) and Clinical Queries Balanced filter (96.9% vs 97.0%; P=.63). Deep learning obtained significantly higher precision than the Clinical Queries Broad filter (34.6% vs 22.4%; P<.001) and McMaster’s textword search (34.6% vs 11.8%; P<.001), but was significantly lower than the Clinical Queries Balanced filter (34.6% vs 40.9%; P<.001). CONCLUSIONS Deep learning performed well compared to state-of-the-art search filters, especially when citations were not indexed. Unlike previous machine learning approaches, the proposed deep learning model does not require feature engineering, or time-sensitive or proprietary features, such as MeSH terms and bibliometrics. Deep learning is a promising approach to identifying reports of scientifically rigorous clinical research. Further work is needed to optimize the deep learning model and to assess generalizability to other areas, such as diagnosis, etiology, and prognosis.

Maps can be a great source of information for a given geographic region, but they can be difficult to find and even harder to process. A significant problem is that many interesting and useful maps are only available in raster format, and even worse many maps have been poorly scanned and they are often compressed with lossy compression algorithms. Furthermore, for many of these maps there is no meta data providing the geographic coordinates, scale, or projection. Previous research on map processing has developed techniques that typically work on maps from a single map source. In contrast, we have developed a general approach to finding and processing street maps. This includes techniques for discovering maps online, extracting geographic and textual features from maps, using the extracted features to determine the geographic coordinates of the maps, and aligning the maps with imagery. The resulting system can find, register, and extract a variety of features from raster maps, which can then be used for various applications, such as annotating satellite imagery, creating and updating maps, or constructing detailed gazetteers.