Hidden Markov models for the stimulus-response relationships of multistate neural systems.

Given recent experimental results suggesting that neural circuits may evolve through multiple firing states, we develop a framework for estimating state-dependent neural response properties from spike train data. We modify the traditional hidden Markov model (HMM) framework to incorporate stimulus-driven, non-Poisson point-process observations. For maximal flexibility, we allow external, time-varying stimuli and the neurons' own spike histories to drive both the spiking behavior in each state and the transitioning behavior between states. We employ an appropriately modified expectation-maximization algorithm to estimate the model parameters. The expectation step is solved by the standard forward-backward algorithm for HMMs. The maximization step reduces to a set of separable concave optimization problems if the model is restricted slightly. We first test our algorithm on simulated data and are able to fully recover the parameters used to generate the data and accurately recapitulate the sequence of hidden states. We then apply our algorithm to a recently published data set in which the observed neuronal ensembles displayed multistate behavior and show that inclusion of spike history information significantly improves the fit of the model. Additionally, we show that a simple reformulation of the state space of the underlying Markov chain allows us to implement a hybrid half-multistate, half-histogram model that may be more appropriate for capturing the complexity of certain data sets than either a simple HMM or a simple peristimulus time histogram model alone.

Accuracy in the prediction and estimation of adherence to bracewear before and during treatment of adolescent idiopathic scoliosis.

BACKGROUND: Because bracing for scoliosis may prevent curve progression, it is important to recognize nonadherence. We used temperature sensors to determine actual bracewear and examined: (1) the ability of a new pretreatment questionnaire to predict bracewear; (2) the ability of the physician and orthotist to predict bracewear before treatment and (3) the ability of physicians, orthotists, patients, and parents to accurately estimate bracewear during the first year of treatment. METHODS: Sixteen males and 108 females with adolescent idiopathic scoliosis were fitted with a Boston brace equipped with a temperature sensor and told that investigators were examining comfort. Before treatment, each patients completed an 18-item Brace-Beliefs Questionnaire (BBQ), and physicians/orthotists rated the likelihood that their patient would be adherent. During treatment, physicians, orthotists, patients, and parents provided estimates of daily bracewear. Data obtained at 1 to 3, 4 to 7, and 9 to 12 months into treatment were analyzed. RESULTS: Scores from the BBQ were related to actual adherence (r = 0.46, P < 0.001). No patient scoring more than 1 SD below the BBQ sample mean had an adherence level more than 40%. Correlations of physician/orthotist pretreatment predictions with actual adherence were minimal. Overall, patients wore the brace 47% of the prescribed time, although they were estimated to have worn it 64%, 66%, 72%, and 75% by physicians, orthotists, parents, and patients, respectively. Physicians/orthotists incorrectly identified at least 1 of every 4 nonadherers. CONCLUSIONS: Predicting a patient's adherence before treatment is difficult, but a pretreatment questionnaire may be helpful. During treatment, all respondents overestimated adherence. Health care providers should be mindful of overreports of bracewear and skeptical of their own assessments of adherence. CLINICAL RELEVANCE: Potential nonadherence may be predicted by a brief treatment-specific questionnaire. Treatment teams should not assume that patients follow their instructions or that family members are accurate sources of adherence information during treatment. Health care providers also should not assume that they can accurately predict adherence based on subjective expectations.