Dataset Information

IntelligentPooling: Practical Thompson Sampling for mHealth.

ABSTRACT: In mobile health (mHealth) smart devices deliver behavioral treatments repeatedly over time to a user with the goal of helping the user adopt and maintain healthy behaviors. Reinforcement learning appears ideal for learning how to optimally make these sequential treatment decisions. However, significant challenges must be overcome before reinforcement learning can be effectively deployed in a mobile healthcare setting. In this work we are concerned with the following challenges: 1) individuals who are in the same context can exhibit differential response to treatments 2) only a limited amount of data is available for learning on any one individual, and 3) non-stationary responses to treatment. To address these challenges we generalize Thompson-Sampling bandit algorithms to develop IntelligentPooling. IntelligentPooling learns personalized treatment policies thus addressing challenge one. To address the second challenge, IntelligentPooling updates each user's degree of personalization while making use of available data on other users to speed up learning. Lastly, IntelligentPooling allows responsivity to vary as a function of a user's time since beginning treatment, thus addressing challenge three.

SUBMITTER: Tomkins S

PROVIDER: S-EPMC8494236 | biostudies-literature | 2021 Sep

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

IntelligentPooling: Practical Thompson Sampling for mHealth.

Tomkins Sabina S Liao Peng P Klasnja Predrag P Murphy Susan S

Machine learning 20210621 9

In mobile health (mHealth) smart devices deliver behavioral treatments repeatedly over time to a user with the goal of helping the user adopt and maintain healthy behaviors. Reinforcement learning appears ideal for learning how to optimally make these sequential treatment decisions. However, significant challenges must be overcome before reinforcement learning can be effectively deployed in a mobile healthcare setting. In this work we are concerned with the following challenges: 1) individuals w ...[more]

PMID: 34621105

Similar Datasets

Project description:Epilepsy is a chronic disorder characterized by recurring seizures, and results from excessive, synchronized discharge in populations of cells. Epilepsy affects 1% of the population and results from a variety of CNS insults. Traumatic physical brain injury causes 4% of cases. Posttraumatic epilepsy (PTE) is characterized by a delay of weeks to years between the trauma and the first seizures. Proper prophylactic intervention during this incubation time might prevent PTE. Many PTE patients are resistant to anticonvulsant medications and suffer from negative side effects. Better knowledge of the cell biology underlying injury-induced epileptogenesis should lead to better therapeutic strategies for interrupting the events occurring between the injury and the development of epilepsy. The aim of my experiments is to identify cellular factors underlying PTE using an in vitro model system to study two key aspects of brain injury in isolation: deafferentation and axonal transection. My aim is to perform a non-biased comparison of the expression of genes in area CA1, and separately in area CA3, in unlesioned cultures and in cultures lesioned 1, 3 and 7 days earlier. Changes in gene expression due to neuronal denervation will be apparent in area CA1, whereas changes in expression due to axonal transection will be apparent in area CA3. In performing these experiments, I hope to gain insight into the molecular underpinnings of the electrophysiological and immunocytochemical effects that we have already detected, as well as to generate new hypotheses that we can test further using our own expertise and techniques. We use cultured slices of rat hippocampus to study the effects of brain injury. Our model is to lesion the Schaffer collateral projection pathway between areas CA1 and CA3, producing a selective injury of the axons of CA3 cells and selective denervation of CA1 cells. Our preliminary electrophysiological and immunocytochemical data indicate that axonal sprouting is triggered in area CA3 and that the dendritic excitability is increased in CA1 pyramidal cells. Both effects occur only after about 7 days, suggesting they are mediated by changes in gene expression. My hypothesis is that the expression of numerous genes changes after Schaffer collateral transection. I predict that neurotrophins and their receptors change in area CA3 to trigger axonal sprouting, along with genes for proteins mediated axonal extension, and that changes in the expression of genes for glutamate receptors and voltage-dependent ion channels occur in area CA1 to account for their increased excitability. Hippocampal slice cultures prepared from male Sprague-Dawley rats at PN day 6. Slices will be maintained in vitro for 14 days in serum-containing medium. Cultures will then be examined and unhealthy cultures discarded. Remaining sister cultures will be divided into two groups: cultures to be lesioned and cultures remaining unlesioned. Lesioned cultures will have a sterile transection of the Schaffer collateral pathway using a razor blade shard. All cultures then returned to the incubator. At various times , cultures will be removed from the incubator. Individual cultures will be soaked in RNAlater buffer(Ambion), and a razor blade shard will be used to subdissect the CA3 and CA1 segments. Segments are sucked up with a sterile pipette-tip containing RNAlater buffer and transferred into a sterile tube. Total RNA will be prepared by homogenization in TRIzol reagent (GibcoBRL) and precipitation in isopropyl alcohol. Rneasy Kit (Qiagen) will further isolate RNA. Spectrographic analysis of RNA prepared indicates 260nm:280nm ratios above 1.6, and insignificant amounts of either degraded RNA or DNA contaminants detected on a 2% denaturing gel. We have found that it is necessary to pool 15 CA3 or CA1 segments to obtain 5+ ug of total RNA. Total RNA samples will be sent to the consortium for additional quality control, labeling, hybridizations using the Affymetrix Neuro U34 chip, scanning, and preliminary analyses. The gene array data will then be compared in several pairwise manners (lesion vs. age matched control and region vs. region, for each time point), as well as time sequence comparisons for gene expression in each region. Keywords: time-course

Project description:BackgroundThe COVID-19 pandemic has become a global public health event, which has raised concerns regarding individuals' health. Individuals need to cope with the COVID-19 pandemic with guidelines on symptom recognition, home isolation, and maintain mental health. Besides routine use of mobile health (mHealth) such as accessing information to keep healthy, individuals can use mHealth services in situations requiring urgent medical care, which is defined as mHealth emergency use. It is not known whether individuals have increased their daily mHealth services emergency use as a result of disruptions caused by the COVID-19 pandemic.ObjectiveThe purpose of this diary analysis study is to assess the influences of daily disruptions related to the COVID-19 pandemic on individuals' mHealth emergency use. The secondary purpose of this study is to explore the mediating role of COVID-19-induced strain and the moderating role of promotion regulatory focus in the relationship between daily disruptions of COVID-19 and mHealth emergency use. Drawing from the cognitive activation theory of stress, we investigated the underlying mechanism and boundary condition of the influence of COVID-19-related disruptions on daily mHealth emergency use.MethodsTo test the proposed model, this study adopts the experience sampling method to collect daily data. The experience sampling method helps researchers to capture participants' fluctuations in emotions, mental engagement in an activity, and experienced stress. This study collected 550 cases nested in 110 samples in mainland China to test the conceptual model. In addition, we employed hierarchical linear modeling analysis to test the effect of COVID-19-related disruptions on mHealth emergency use.ResultsWe found that COVID-19-related disruptions increased COVID-19-induced strain (γ=0.24, P<.001) and mHealth emergency use on a daily basis (γ=0.28, P<.001). COVID-19-induced daily strain mediated this relationship (effect=0.09, 95% CI 0.05-0.14). Promotion regulatory focus moderated the relationship between COVID-19-induced strain and mHealth emergency use (γ=0.35, P=.02). In addition, the indirect relationship between disruptions and mHealth emergency use intentions through COVID-19-induced strain is contingent upon promotion regulatory focus: this relationship was stronger in those with high promotion regulatory focus (effect=0.12, 95% CI 0.06-0.19) than in those with low promotion regulatory focus (effect=0.06, 95% CI 0.02-0.11).ConclusionsEvent disruption of the COVID-19 pandemic induced mHealth emergency use intention through increased psychological strain. Furthermore, individuals' promotion regulatory focus amplified this indirect relationship. Our findings extend our understanding of the factors underlying mHealth emergency use intention and illustrate the potential contingent role of promotion regulatory focus in the cognitive activation theory of stress. This study also opens avenues for future research on mHealth emergency use intention in other countries and cultural settings.

Dataset Information

IntelligentPooling: Practical Thompson Sampling for mHealth.

Publications

IntelligentPooling: Practical Thompson Sampling for mHealth.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets