bioRxiv Science⌕ Search

Biology subjects

Tung, M.

Publications and source records attributed to Tung, M..

2 recordsLinked to original sources

Spatial autocorrelation dimension as a potential determinant for the temporal persistence of human atrial and ventricular fibrillation.

Background: Despite being central to atrial fibrillation (AF) and ventricular fibrillation (VF) mechanisms and therapy, the factors governing AF and VF termination are poorly understood. It has been noted that ratio of system size (L) and the two-point spatial correlation length ({xi}2) are associated with time until termination in transient spatiotemporally chaotic systems, but the relationship between these characteristics and termination has not been systematically studied in human AF and VF. Objective: We aimed assess whether the time to cardiac fibrillation termination can be predicted using a novel estimator, the spatial autocorrelation dimension (Di), defined as the ratio of L and {xi}2, in human AF and VF. Methods: Di was computed and compared in a multi-centre, multi-system study with data for sustained versus spontaneously terminating human AF/VF. VF data was collected during coronary-bypass surgery; and AF data during clinically indicated AF ablation. We analyzed: i) VF mapped using a 256-electrode epicardial sock (n=12pts); ii) AF mapped using a 64-electrode constellation basket-catheter (n=15pts); iii) AF mapped using a 16-electrode HD-grid catheter (n=42pts). To investigate temporal fibrillation persistence, the response of AF-episodes to flecainide (n=7pts) was also studied. Results: Spontaneously terminating fibrillation demonstrated a lower Di (P<0.001 all systems). Lower Di was also seen in paroxysmal compared to persistent AF (P=0.002). Post-flecainide, Di decreased over time (P<0.001). Lower Di was also associated with longer-lasting episodes of AF/VF (R2>0.90, P<0.05 in all cases). Using k-means clustering, two distinct clusters and their centroids were identified i) a cluster of spontaneously terminating episodes, and ii) a cluster of sustained epochs. Conclusion: Di predicts the temporal persistence of cardiac fibrillation. This finding provides potentially important insights into a possible common pathway to termination and therapeutic approaches.

physiology↗

MatchMiner: An open source computational platform for real-time matching of cancer patients to precision medicine clinical trials using genomic and clinical criteria

BackgroundMolecular profiling of cancers is now routine at many cancer centers, and the number of precision cancer medicine clinical trials, which are informed by profiling, is steadily rising. Additionally, these trials are becoming increasingly complex, often having multiple arms and many genomic eligibility criteria. Currently, it is a challenging for physicians to match patients to relevant clinical trials using the patients genomic profile, which can lead to missed opportunities. Automated matching against uniformly structured and encoded genomic eligibility criteria is essential to keep pace with the complex landscape of precision medicine clinical trials.\n\nResultsTo meet these needs, we built and deployed an automated clinical trial matching platform called MatchMiner at the Dana-Farber Cancer Institute (DFCI). The platform has been integrated with Profile, DFCIs enterprise genomic profiling project, which contains tumor profile data for >20,000 patients, and has been made available to physicians across the Institute. As no current standard exists for encoding clinical trial eligibility criteria, a new language called Clinical Trial Markup Language (CTML) was developed, and over 178 genomically-driven clinical trials were encoded using this language. The platform is open source and freely available for adoption by other institutions.\n\nConclusionMatchMiner is the first open platform developed to enable computational matching of patient-specific genomic profiles to precision cancer medicine clinical trials. Creating MatchMiner required developing clinical trial eligibility standards to support genome-driven matching and developing intuitive interfaces to support practical use-cases. Given the complexity of tumor profiling and the rapidly changing multi-site nature of genome-driven clinical trials, open source software is the most efficient, scalable, and economical option for matching cancer patients to clinical trials.

bioinformatics↗