Search

Background: Despite the effectiveness of levodopa for treatment of Parkinson's disease (PD), prolonged usage leads to development of motor complications, most notably levodopa-induced dyskinesia (LID). Persons with PD and their physicians must regularly modify treatment regimens and timing for optimal relief of symptoms. While standardized clinical rating scales exist for assessing the severity of PD symptoms, they must be administered by a trained medical professional and are inherently subjective. Computer vision is an attractive, non-contact, potential solution for automated assessment of PD, made possible by recent advances in computational power and deep learning algorithms. The objective of this paper was to evaluate the feasibility of vision-based assessment of parkinsonism and LID using pose estimation.Methods: Nine participants with PD and LID completed a levodopa infusion protocol, where symptoms were assessed at regular intervals using the Unified Dyskinesia Rating Scale (UDysRS) and Unified Parkinson's Disease Rating Scale (UPDRS). Movement trajectories of individual joints were extracted from videos of PD assessment using Convolutional Pose Machines, a pose estimation algorithm built with deep learning. Features of the movement trajectories (e.g. kinematic, frequency) were used to train random forests to detect and estimate the severity of parkinsonism and LID. Communication and drinking tasks were used to assess LID, while leg agility and toe tapping tasks were used to assess parkinsonism. Feature sets from tasks were also combined to predict total UDysRS and UPDRS Part III scores.Results: For LID, the communication task yielded the best results (detection: AUC = 0.930, severity estimation: r = 0.661). For parkinsonism, leg agility had better results for severity estimation (r = 0.618), while toe tapping was better for detection (AUC = 0.773). UDysRS and UPDRS Part III scores were predicted with r = 0.741 and 0.530, respectively.Conclusion: The proposed system provides insight into the potential of computer vision and deep learning for clinical application in PD and demonstrates promising performance for the future translation of deep learning to PD clinical practices. Convenient and objective assessment of PD symptoms will facilitate more frequent touchpoints between patients and clinicians, leading to better tailoring of treatment and quality of care. [ABSTRACT FROM AUTHOR]

Background: Clustering time-series data into discrete groups can improve prediction and provide insight into the nature of underlying, unobservable states of the system. However, temporal variation in probabilities of group occupancy, or the rates at which individuals move between groups, can obscure such signals. We use finite mixture and hidden Markov models (HMMs), two standard clustering techniques, to model long-term hourly movement data from Florida panthers ( Puma concolor coryi ). Allowing for temporal heterogeneity in transition probabilities, a straightforward but little-used extension of the standard HMM framework, resolves some shortcomings of current models and clarifies the movement patterns of panthers., Results: Simulations and analyses of panther data showed that model misspecification (omitting important sources of variation) can lead to overfitting/overestimating the underlying number of movement states. Models incorporating temporal heterogeneity identify fewer underlying states, and can make out-of-sample predictions that capture observed diurnal and autocorrelated movement patterns exhibited by Florida panthers., Conclusion: Incorporating temporal heterogeneity improved goodness of fit and predictive capability as well as reducing the selected number of movement states closer to a biologically interpretable level, although there is further room for improvement here. Our results suggest that incorporating additional structure in statistical models of movement can allow more accurate assessment of appropriate model complexity.

Background: Discovery of new medicinal agents from natural sources has largely been an adventitious process based on screening of plant and microbial extracts combined with bioassay-guided identification and natural product structure elucidation. Increasingly rapid and more cost-effective genome sequencing technologies coupled with advanced computational power have converged to transform this trend toward a more rational and predictive pursuit., Results: We have developed a rapid method of scanning genome sequences for multiple polyketide, nonribosomal peptide, and mixed combination natural products with output in a text format that can be readily converted to two and three dimensional structures using conventional software. Our open-source and web-based program can assemble various small molecules composed of twenty standard amino acids and twenty two other chain-elongation intermediates used in nonribosomal peptide systems, and four acyl-CoA extender units incorporated into polyketides by reading a hidden Markov model of DNA. This process evaluates and selects the substrate specificities along the assembly line of nonribosomal synthetases and modular polyketide synthases., Conclusion: Using this approach we have predicted the structures of natural products from a diverse range of bacteria based on a limited number of signature sequences. In accelerating direct DNA to metabolomic analysis, this method bridges the interface between chemists and biologists and enables rapid scanning for compounds with potential therapeutic value.