2017/11/11 by Avinash Bukkittu, Baihan Lin, Bukkittu, Avinash +7
Computer Science · Neuroscience · #Artificial Intelligence (cs.AI) #FOS: Biological sciences #FOS: Computer and information sciences #Machine Learning (stat.ML) #Music Technology and Sound Studies #Music and Audio Processing #Neural dynamics and brain function #Quantitative Methods (q-bio.QM)
paper · pdf · doi:10.48550/arxiv.1711.04078
openalex publication_date 2017/11/11 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
We search for digital biomarkers from Parkinson's Disease by observing\napproximate repetitive patterns matching hypothesized step and stride periodic\ncycles. These observations were modeled as a cycle of hidden states with\nrandomness allowing deviation from a canonical pattern of transitions and\nemissions, under the hypothesis that the averaged features of hidden states\nwould serve to informatively characterize classes of patients/controls. We\npropose a Hidden Semi-Markov Model (HSMM), a latent-state model, emitting\n3D-acceleration vectors. Transitions and emissions are inferred from data. We\nfit separate models per unique device and training label. Hidden Markov Models\n(HMM) force geometric distributions of the duration spent at each state before\ntransition to a new state. Instead, our HSMM allows us to specify the\ndistribution of state duration. This modified version is more effective because\nwe are interested more in each state's duration than the sequence of distinct\nstates, allowing inclusion of these durations the feature vector.\n