From Group-Level Statistics to Single-Subject Prediction: Machine Learning Detection of Concussion in Retired Athletes

2019 ◽  
Vol 27 (7) ◽  
pp. 1492-1501 ◽  
Author(s):  
Rober Boshra ◽  
Kiret Dhindsa ◽  
Omar Boursalie ◽  
Kyle I. Ruiter ◽  
Ranil Sonnadara ◽  
...  
Keyword(s):  
Machine Learning ◽  
Single Subject ◽  
Group Level ◽  
Level Statistics

scholarly journals Using a machine learning approach to complement group level statistics in experimental psychology: A case study to reveal different levels of inhibition in a modified Flanker Task

2018 ◽  
Author(s):  
Tanja Krumpe ◽  
Christian Scharinger ◽  
Wolfgang Rosenstiel ◽  
Peter Gerjets ◽  
Martin Spueler
Keyword(s):  
Machine Learning ◽  
Executive Function ◽  
Inhibitory Control ◽  
Flanker Task ◽  
Mental States ◽  
Support Vector ◽  
Group Level ◽  
Single Trial ◽  
Level Statistics ◽  
Different Levels

In this paper, we demonstrate how machine learning (ML) can be used to beneficially complement the traditional analysis of behavioral and physiological data to provide new insights into the structure of mental states, in this case, executive functions (EFs) with a focus on inhibitory control. We used a modified Flanker task with the aim to distinguish three levels of inhibitory control: no inhibition, readiness for inhibition and the actual execution of inhibitory control. A simultaneously presented n-back task was used to additionally induce demands on a second executive function. This design enabled us to investigate how the overlap of resources influences the distinction between three levels of inhibitory control. A support vector machine (SVM) based classification approach has been used on EEG data to predict the level of inhibitory control on single-subject and single-trial level. The SVM classification is a subject-specific and single-trial based approach which will be compared to standard group-level statistical approaches to reveal that both methodologies access different properties of the data. We show that considering both methods can give new insights into mental states which cannot be discovered when only using group-level statistics alone. Machine learning results indicate that three different levels of inhibitory control can be distinguished, while the group-average analysis does not give rise to this assumption. In addition, we highlight one other important benefit of the ML approach. We are able to define specific properties of the executive function inhibition by investigating the neural activation patterns that were used during the classification process.

scholarly journals Predicting Future Service Use in Dutch Mental Healthcare: A Machine Learning Approach

2021 ◽  
Author(s):  
Kasper van Mens ◽  
Sascha Kwakernaak ◽  
Richard Janssen ◽  
Wiepke Cahn ◽  
Joran Lokkerbol ◽  
...  
Keyword(s):  
Machine Learning ◽  
Service Use ◽  
Visual Analysis ◽  
Mental Healthcare ◽  
Machine Learning Techniques ◽  
Individual Treatment ◽  
Learning Approach ◽  
Group Level ◽  
Starting Point ◽  
Machine Learning Approach

AbstractA mental healthcare system in which the scarce resources are equitably and efficiently allocated, benefits from a predictive model about expected service use. The skewness in service use is a challenge for such models. In this study, we applied a machine learning approach to forecast expected service use, as a starting point for agreements between financiers and suppliers of mental healthcare. This study used administrative data from a large mental healthcare organization in the Netherlands. A training set was selected using records from 2017 (N = 10,911), and a test set was selected using records from 2018 (N = 10,201). A baseline model and three random forest models were created from different types of input data to predict (the remainder of) numeric individual treatment hours. A visual analysis was performed on the individual predictions. Patients consumed 62 h of mental healthcare on average in 2018. The model that best predicted service use had a mean error of 21 min at the insurance group level and an average absolute error of 28 h at the patient level. There was a systematic under prediction of service use for high service use patients. The application of machine learning techniques on mental healthcare data is useful for predicting expected service on group level. The results indicate that these models could support financiers and suppliers of healthcare in the planning and allocation of resources. Nevertheless, uncertainty in the prediction of high-cost patients remains a challenge.

scholarly journals Imaging data reveal divergent longitudinal trajectories in ALS, PLS and poliomyelitis: group-level and single-subject traits

Data in Brief ◽  
2021 ◽  
pp. 107484
Author(s):  
Marlene Tahedl ◽  
Stacey Li Hi Shing ◽  
Eoin Finegan ◽  
Rangariroyashe H. Chipika ◽  
Jasmin Lope ◽  
...  
Keyword(s):  
Single Subject ◽  
Group Level ◽  
Imaging Data ◽  
Longitudinal Trajectories

Treatment of Underlying Forms: A Bayesian Meta-Analysis of the Effects of Treatment and Person-Related Variables on Treatment Response

2021 ◽  
pp. 1-21
Author(s):  
Alexander M. Swiderski ◽  
Yina M. Quique ◽  
Michael Walsh Dickey ◽  
William D. Hula
Keyword(s):  
Sentence Comprehension ◽  
Meta Analysis ◽  
Interrupted Time Series ◽  
Single Subject ◽  
Group Level ◽  
Sentence Level ◽  
Hypothesis Model ◽  
Treatment Dose ◽  
Effect Of Treatment ◽  
Generalized Linear Mixed Effects

Purpose This meta-analysis synthesizes published studies using “treatment of underlying forms” (TUF) for sentence-level deficits in people with aphasia (PWA). The study aims were to examine group-level evidence for TUF efficacy, to characterize the effects of treatment-related variables (sentence structural family and complexity; treatment dose) in relation to the Complexity Account of Treatment Efficacy (CATE) hypothesis, and to examine the effects of person-level variables (aphasia severity, sentence comprehension impairment, and time postonset of aphasia) on TUF response. Method Data from 13 single-subject, multiple-baseline TUF studies, including 46 PWA, were analyzed. Bayesian generalized linear mixed-effects interrupted time series models were used to assess the effect of treatment-related variables on probe accuracy during baseline and treatment. The moderating influence of person-level variables on TUF response was also investigated. Results The results provide group-level evidence for TUF efficacy demonstrating increased probe accuracy during treatment compared with baseline phases. Greater amounts of TUF were associated with larger increases in accuracy, with greater gains for treated than untreated sentences. The findings revealed generalization effects for sentences that were of the same family but less complex than treated sentences. Aphasia severity may moderate TUF response, with people with milder aphasia demonstrating greater gains compared with people with more severe aphasia. Sentence comprehension performance did not moderate TUF response. Greater time postonset of aphasia was associated with smaller improvements for treated sentences but not for untreated sentences. Conclusions Our results provide generalizable group-level evidence of TUF efficacy. Treatment and generalization responses were consistent with the CATE hypothesis. Model results also identified person-level moderators of TUF (aphasia severity, time postonset of aphasia) and preliminary estimates of the effects of varying amounts of TUF for treated and untreated sentences. Taken together, these findings add to the TUF evidence and may guide future TUF treatment–candidate selection. Supplemental Material https://doi.org/10.23641/asha.16828630

scholarly journals Preprocessed Consortium for Neuropsychiatric Phenomics dataset

F1000Research ◽  
2017 ◽  
Vol 6 ◽  
pp. 1262 ◽  
Author(s):  
Krzysztof J. Gorgolewski ◽  
Joke Durnez ◽  
Russell A. Poldrack
Keyword(s):  
Single Subject ◽  
Group Level ◽  
Original Dataset ◽  
Potential Confound

Here we present preprocessed MRI data of 265 participants from the Consortium for Neuropsychiatric Phenomics (CNP) dataset. The preprocessed dataset includes minimally preprocessed data in the native, MNI and surface spaces accompanied with potential confound regressors, tissue probability masks, brain masks and transformations. In addition the preprocessed dataset includes unthresholded group level and single subject statistical maps from all tasks included in the original dataset. We hope that availability of this dataset will greatly accelerate research.

scholarly journals Doing nothing and what it looks like: inactivity in fattening cattle

PeerJ ◽  
2020 ◽  
Vol 8 ◽  
pp. e9395
Author(s):  
Sara Hintze ◽  
Freija Maulbetsch ◽  
Lucy Asher ◽  
Christoph Winckler
Keyword(s):  
Machine Learning ◽  
Learning Algorithm ◽  
Statistical Hypothesis ◽  
Machine Learning Algorithm ◽  
Body Parts ◽  
Group Level ◽  
Statistical Hypothesis Testing ◽  
Level Data ◽  
Eyes Open ◽  
Husbandry System

Background Animals kept in barren environments often show increased levels of inactivity and first studies indicate that inactive behaviour may reflect boredom or depression-like states. However, to date, knowledge of what inactivity looks like in different species is scarce and methods to precisely describe and analyse inactive behaviour are thus warranted. Methods We developed an Inactivity Ethogram including detailed information on the postures of different body parts (Standing/Lying, Head, Ears, Eyes, Tail) for fattening cattle, a farm animal category often kept in barren environments. The Inactivity Ethogram was applied to Austrian Fleckvieh heifers kept in intensive, semi-intensive and pasture-based husbandry systems to record inactive behaviour in a range of different contexts. Three farms per husbandry system were visited twice; once in the morning and once in the afternoon to cover most of the daylight hours. During each visit, 16 focal animals were continuously observed for 15 minutes each (96 heifers per husbandry system, 288 in total). Moreover, the focal animals’ groups were video recorded to later determine inactivity on the group level. Since our study was explorative in nature, we refrained from statistical hypothesis testing, but analysed both the individual- and group-level data descriptively. Moreover, simultaneous occurrences of postures of different body parts (Standing/Lying, Head, Ears and Eyes) were analysed using the machine learning algorithm cspade to provide insight into co-occurring postures of inactivity. Results Inspection of graphs indicated that with increasing intensity of the husbandry system, more animals were inactive (group-level data) and the time the focal animals were inactive increased (individual-level data). Frequently co-occurring postures were generally similar between husbandry systems, but with subtle differences. The most frequently observed combination on farms with intensive and semi-intensive systems was lying with head up, ears backwards and eyes open whereas on pasture it was standing with head up, ears forwards and eyes open. Conclusion Our study is the first to explore inactive behaviour in cattle by applying a detailed description of postures from an Inactivity Ethogram and by using the machine learning algorithm cspade to identify frequently co-occurring posture combinations. Both the ethogram created in this study and the cspade algorithm may be valuable tools in future studies aiming to better understand different forms of inactivity and how they are associated with different affective states.

scholarly journals ENIGMA HALFpipe: Interactive, reproducible, and efficient analysis for resting-state and task-based fMRI data

2021 ◽  
Author(s):  
Lea Waller ◽  
Susanne Erk ◽  
Elena Pozzi ◽  
Yara J. Toenders ◽  
Courtney C. Haswell ◽  
...  
Keyword(s):  
Quality Assessment ◽  
Resting State ◽  
Low Frequency ◽  
Fmri Data ◽  
Single Subject ◽  
Major Advance ◽  
Group Level ◽  
Imaging Data ◽  
Post Processing ◽  
Individual Subject

The reproducibility crisis in neuroimaging has led to an increased demand for standardized data processing workflows. Within the ENIGMA consortium, we developed HALFpipe (Harmonized AnaLysis of Functional MRI pipeline), an open-source, containerized, user-friendly tool that facilitates reproducible analysis of task-based and resting-state fMRI data through uniform application of preprocessing, quality assessment, single-subject feature extraction, and group-level statistics. It provides state-of-the-art preprocessing using fMRIPrep without the requirement for input data in Brain Imaging Data Structure (BIDS) format. HALFpipe extends the functionality of fMRIPrep with additional preprocessing steps, which include spatial smoothing, grand mean scaling, temporal filtering, and confound regression. HALFpipe generates an interactive quality assessment (QA) webpage to assess the quality of key preprocessing outputs and raw data in general. HALFpipe features myriad post-processing functions at the individual subject level, including calculation of task-based activation, seed-based connectivity, network-template (or dual) regression, atlas-based functional connectivity matrices, regional homogeneity (ReHo), and fractional amplitude of low frequency fluctuations (fALFF), offering support to evaluate a combinatorial number of features or preprocessing settings in one run. Finally, flexible factorial models can be defined for mixed-effects regression analysis at the group level, including multiple comparison correction. Here, we introduce the theoretical framework in which HALFpipe was developed, and present an overview of the main functions of the pipeline. HALFpipe offers the scientific community a major advance toward addressing the reproducibility crisis in neuroimaging, providing a workflow that encompasses preprocessing, post-processing, and QA of fMRI data, while broadening core principles of data analysis for producing reproducible results. Instructions and code can be found at https://github.com/HALFpipe/HALFpipe.

scholarly journals EXPERIMENTAL MACHINE LEARNING OF FINGER PHOTOPLETHYSMOGRAPHY (PPG) FOR AUTONOMOUS HOSPITAL BED PUSHING FRAMEWORK USING POLYNOMIAL REGRESSION

2020 ◽  
Vol 5 (9) ◽  
pp. 108-119
Author(s):  
Yan Hao Tan ◽  
Holden Li King Ho
Keyword(s):  
Machine Learning ◽  
Polynomial Regression ◽  
Influencing Factor ◽  
Life Safety ◽  
Single Subject ◽  
Single Male ◽  
Skilled Nursing ◽  
Non Invasive ◽  
Hospital Bed ◽  
Validation Score

In community-based healthcare, the nursing workforce requires low-skilled nursing automation in the hospital to accelerate talent development towards high-skilled advance practice nurse for community deployment. As precursor, the hospital bed pushing operation for medium-risk patient was hypothesized as a novice nursing task where artificial intelligence automation is possible. The solution framework was embodied by a concept of operation with non-invasive vitals monitoring as priority to study feasibility in addressing patient life-safety requirements. Polynomial regression machine learning of 65 one-hour sets of finger PPG data from a single subject were collected and studied. Convergence of finger PPG to 8th degree polynomial was observed which suggested process feasibility towards establishing patient safe states during autonomous journey. Process reliability ranged between 2% to 95% with long PPG counts as influencing factor for drops in reliability score. Motivation/Background: A predictable non-invasive vitals monitoring was priority to enable autonomous hospital bed pushing framework to address patient life-safety concerns during autonomous journey. Finger PPG is a non-invasive and easy to use method to monitor heart related activities and used to study for convergence and reliability within the framework. Method:65 one-hour sets of finger PPG was recorded from a single male, age 27 subject. The data was processed by polynomial regression machine learning technique to output the degree of polynomial with highest cross validation score mean. Results: Convergence of regressed PPG data to 8th degree for both pre-journey and journey datasets and degree of polynomial matching reliability of 2% to 95% were observed. Conclusions: Convergence of PPG data facilitates the establishment of safe physical states in vitals monitoring, enabling the autonomous hospital bed pushing framework for further development. Reliability remains an area for improvement via medical grade.

scholarly journals Predictability-Based Source Segregation and Sensory Deviance Detection in Auditory Aging

2021 ◽  
Vol 15 ◽  
Author(s):  
Christiane R. Neubert ◽  
Alexander P. Förstel ◽  
Stefan Debener ◽  
Alexandra Bendixen
Keyword(s):  
Sound Source ◽  
Predictive Coding ◽  
Single Subject ◽  
Group Level ◽  
Sound Sources ◽  
Passive Listening ◽  
Deviance Detection ◽  
Sound Source Segregation ◽  
High Degree ◽  
Source Segregation

When multiple sound sources are present at the same time, auditory perception is often challenged with disentangling the resulting mixture and focusing attention on the target source. It has been repeatedly demonstrated that background (distractor) sound sources are easier to ignore when their spectrotemporal signature is predictable. Prior evidence suggests that this ability to exploit predictability for foreground-background segregation degrades with age. On a theoretical level, this has been related with an impairment in elderly adults’ capabilities to detect certain types of sensory deviance in unattended sound sequences. Yet the link between those two capacities, deviance detection and predictability-based sound source segregation, has not been empirically demonstrated. Here we report on a combined behavioral-EEG study investigating the ability of elderly listeners (60–75 years of age) to use predictability as a cue for sound source segregation, as well as their sensory deviance detection capacities. Listeners performed a detection task on a target stream that can only be solved when a concurrent distractor stream is successfully ignored. We contrast two conditions whose distractor streams differ in their predictability. The ability to benefit from predictability was operationalized as performance difference between the two conditions. Results show that elderly listeners can use predictability for sound source segregation at group level, yet with a high degree of inter-individual variation in this ability. In a further, passive-listening control condition, we measured correlates of deviance detection in the event-related brain potential (ERP) elicited by occasional deviations from the same spectrotemporal pattern as used for the predictable distractor sequence during the behavioral task. ERP results confirmed neural signatures of deviance detection in terms of mismatch negativity (MMN) at group level. Correlation analyses at single-subject level provide no evidence for the hypothesis that deviance detection ability (measured by MMN amplitude) is related to the ability to benefit from predictability for sound source segregation. These results are discussed in the frameworks of sensory deviance detection and predictive coding.

Sign in / Sign up

Export Citation Format

Share Document