Physical activity and the outcome of cognitive trajectory: a machine learning approach

Classification and prediction tasks are common in health research. With the increasing availability of vast health data repositories (e.g. electronic medical record databases) and advances in computing power, traditional statistical approaches are being augmented or replaced with machine learning (ML) approaches to classify and predict health outcomes. ML describes the automated process of identifying ("learning") patterns in data to perform tasks. Developing an ML model includes selecting between many ML models (e.g. decision trees, support vector machines, neural networks); model specifications such as hyperparameter tuning; and evaluation of model performance. This process is conducted repeatedly to find the model and corresponding specifications that optimize some measure of model performance. ML models can make more accurate classifications and predictions than their statistical counterparts and confer greater flexibility when modelling unstructured data or interactions between covariates; however, many ML models require larger sample sizes to achieve good classification or predictive performance and have been criticized as "black box" for their poor transparency and interpretability. ML holds potential in family medicine for risk profiling of patients' disease risk and clinical decision support to present additional information at times of uncertainty or high demand. In the future, ML approaches are positioned to become commonplace in family medicine. As such, it is important to understand the objectives that can be addressed using ML approaches and the associated techniques and limitations. This article provides a brief introduction into the use of ML approaches for classification and prediction tasks in family medicine.

BackgroundThe evidence of social robots for people with dementia (PwD) living at home is unclear. The aim was to explore the effects of Coach Pepper (socially assistive humanoid robot system “AMIGO” combined with external tablet‐based training) versus an exclusively tablet‐based training on quality of life, motivation, cognition, mobility, care dependency, depression, and neuropsychiatric symptoms, as well as the user experience with Coach Pepper.MethodsThe study (2019‐2020) includes a randomized controlled trial with complementary interviews using content analysis. The intervention group (16 PwD) received Coach Pepper and the control group (16 PwD) an exclusively tablet‐based training for three weeks. Data were collected using standardized questionnaires (AES, MOCA, DEMQOL, PAS, TUG, GDS, NPI) and interviews.ResultsIn both groups, PwD were >80 years old and mainly female. The mild stage of Alzheimer's was predominant in both groups. The comparison of quality of life (DEMQOL questionnaire) between the groups showed a significantly greater increase in the Coach Pepper group (X̅: 5.48 vs. ‐0.31; med: 4 vs. 2 points, p = 0.0226). Furthermore, global quality of life increased significantly in the Coach Pepper group (X̅: 0.4 vs. ‐0.2; med: 1.0 vs. 0.0 points; p = 0.013). All other outcomes did not differ significantly. Main qualitative results: Attitude/feelings ‐ PwD reacted mainly positively to Pepper. The PwD were generally not afraid of Pepper and did not feel repulsed. They found Pepper nice/pleasant and funny. Communication, social contacts ‐ Pepper was received well and integrated as a family member. His presence was not disturbing, more an enrichment. PwD regarded Pepper as a friend, child and contact person. He promoted communication in PwD. The life of PwD “bloomed” and they bonded with Pepper, which caused feelings of sadness after Pepper moved out. Meaningful activities‐ music and dance activities stimulated PwD particularly positively.ConclusionCoach Pepper significantly increased the quality of life of the PwD in the robot group. As the additional qualitative results demonstrate, this increase can be attributed in particular to the promotion of communication and social contacts by Pepper. Further research should focus on the increase of Pepper’s communication skills and program content.

Introduction Ensuring equitable performance of sleep-related machine learning (ML) models is vital for public health and health equity. This research sought to determine the impact of sociodemographic and health disparities factors on the performance and characteristics of ML models designed to predict OSA treatment initiation. Methods Our data source was the All-Payer Claims Database (APCD) for the Wisconsin Health Information Organization (WHIO). Inclusion criteria included continuous insurance coverage for >12 months prior and >30 months after OSA diagnosis (defined by OCD code G47.33), and having undergone OSA diagnostic testing (defined by CPT codes). OSA treatment was defined based on durable medical equipment charges for PAP machines and supplies. Sociodemographic variables were extracted from APCD and included race, gender, age, and area socioeconomic deprivation (Area Deprivation Index; ADI). The ADI is a validated marker of health risk based on 17 health disparities factors ranking relative disadvantage across communities. Random Forest ML models were trained to predict OSA treatment initiation using sociodemographic variables and a medication history including 39,712 unique medications codified across 94 medication categories. Results Of N=6,026,463 subjects in the ACPD, n=154,821 underwent OSA diagnostic testing, and n=43,601 were diagnosed with OSA. 10-Fold Cross-Validation training-testing was applied to estimate sensitivity-specificity of ML models for predicting treatment initiation. Receiver operating characteristic curve area-under-the-curve (ROC-AUC) analyses were used to compare relative differences in predictive power of each variable. In ROC-AUC analysis of individual variables, the power for predicting OSA treatment were observed in relative rank order by age (0.568), race (0.547), ADI national-level (0.545), ADI state-level (0.544), gender (0.524), and medication history (0.516). In ROC-AUC analysis of combination variables, the highest ML model performance was observed in the combination of only three-variables (age-gender-race, 0.600), while the combination of all-variables showed an ROC-AUC value of (0.594), and the resulting performance difference was not statistically significant based on comparison of the ROC-AUC measures. Conclusion Demographic and health disparity factors may play an important role in future development of predictive AI/ML models. Population sleep health data represents an important resource to identify and bridge care gaps, reduce sleep health disparities, and achieve health equity. Support (if any)

We previously developed machine learning (ML) models for predicting cycloplegic spherical equivalent refraction (SER) and myopia using non-cycloplegic data and following a standardized protocol (cycloplegia with 0.5% tropicamide and biometry using NIDEK A-scan), but the models’ performance may not be generalizable to other settings. This study evaluated the performance of ML models in an independent cohort using a different cycloplegic agent and biometer. Chinese students (N = 614) aged 8–13 years underwent autorefraction before and after cycloplegia with 0.5% tropicamide (n = 505) or 1% cyclopentolate (n = 109). Biometric measures were obtained using an IOLMaster 700 (n = 207) or Optical Biometer SW-9000 (n = 407). ML models were evaluated using R2, mean absolute error (MAE), sensitivity, specificity, and area under the ROC curve (AUC). The XGBoost model predicted cycloplegic SER very well (R2 = 0.95, MAE (SD) = 0.32 (0.30) D). Both ML models predicted myopia well (random forest: AUC 0.99, sensitivity 93.7%, specificity 96.4%; XGBoost: sensitivity 90.1%, specificity 96.8%) and accurately predicted the myopia rate (observed 62.9%; random forest: 60.6%; XGBoost: 58.8%) despite heterogeneous cycloplegia and biometry factors. In this independent cohort of students, XGBoost and random forest performed very well for predicting cycloplegic SER and myopia status using non-cycloplegic data. This external validation study demonstrated that ML may provide a useful tool for estimating cycloplegic SER and myopia prevalence with heterogeneous clinical parameters, and study in additional populations is warranted.

BackgroundSocial robots are anthropomorphised platforms developed to interact with humans, using natural language, offering an accessible and intuitive interface suited to diverse cognitive abilities. Social robots can be used to support people with dementia (PwD) and carers in their homes managing medication, hydration, appointments, and evaluating mood, wellbeing, and potentially cognitive decline. Such robots have potential to reduce care burden and prolong independent living, yet translation into PwD use remains insignificant.MethodWe have developed two social robots ‐ a conversational robot and a digital social robot for mobile devices capable of communicating through natural language (powered by Amazon Alexa) and facial expressions that ask PwD daily questions about their health and wellbeing and also provide digital assistant functionality. We record data comprising of PwD’s responses to daily questions, audio speech and text of conversations with Alexa to automatically monitor their health and wellbeing using machine learning. We followed user‐centric development processes by conducting focus groups with 13 carers, 2 PwD and 5 clinicians to iterate the design. We are testing social robot with 3 PwD in their homes for ten weeks.ResultWe received positive feedback on social robot from focus group participants. Ease of use, low maintenance, accessibility, assistance with medication, supporting with health and wellbeing were identified as the key opportunities for social robots. Based on responses to a daily questionnaire, our robots generate a report detailing PwD wellbeing that is automatically sent via email to family members or carers. This information is also stored systematically in a database that can help clinicians monitor their patients remotely. We use natural language processing to analyse conversations and identify topics of interest to PwD such that robot behaviour could be adapted. We process speech using signal processing and machine learning classifiers to identify emotions and mood.ConclusionAffective social robots with voice and facial expressions have a great potential for enhancing dementia care at home. Combining responses to daily questions, topics of conversations and emotions from speech analysis can help assess daily status of PwD automatically and monitor their health and well‐being over time remotely.

To develop and internally validate a machine learning (ML) model that identifies older outpatients with MCI using routine electronic health record (EHR) data. We conducted a retrospective cross-sectional study of community outpatients aged ≥60 years in Zhejiang, China. Structured EHR predictors included demographics, comorbidities/medications, lifestyle, and visit patterns. The outcome was adjudicated MCI based on cognitive screening (MoCA plus supplemental tests). Supervised ML classifiers were compared using 10-fold cross-validation and an independent held-out test set; class imbalance was addressed with SMOTE. Performance was assessed by the area under the ROC curve (AUC) and by sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), accuracy, and F1 score. The test set included ~640 patients (≈20% MCI). Gradient Boosting performed best: cross-validation mean AUC 0.855 (SD 0.031) and accuracy 0.862 (SD 0.013); test AUC 0.850, accuracy 0.833, and F1 0.402. At the default threshold, sensitivity was 0.286 and specificity 0.967 (PPV 0.679; NPV 0.847). Prioritizing sensitivity (~0.82) lowered specificity (~0.64). At a high-sensitivity threshold of 0.159, the model achieved a sensitivity of 0.802 with a specificity of 0.751 (PPV 0.441; NPV 0.939). Important predictors included older age, female sex, lower education, smaller family size, and higher depression scores. An ML model using routine outpatient EHR can discriminate MCI in older adults (AUC ≈ 0.85), supporting potential for automated, low-cost screening in primary care. Using the predicted probabilities generated in this analysis, we assessed calibration and conducted a decision-curve analysis. While the model shows good discrimination and calibration, external validation is still required to confirm clinical utility and refine operating thresholds.

To construct and validate the best predictive model for 28-day death risk in patients with septic shock based on different supervised machine learning algorithms. The patients with septic shock meeting the Sepsis-3 criteria were selected from Medical Information Mart for Intensive Care-IV v2.0 (MIMIC-IV v2.0). According to the principle of random allocation, 70% of these patients were used as the training set, and 30% as the validation set. Relevant predictive variables were extracted from three aspects: demographic characteristics and basic vital signs, serum indicators within 24 hours of intensive care unit (ICU) admission and complications possibly affecting indicators, functional scoring and advanced life support. The predictive efficacy of models constructed using five mainstream machine learning algorithms including decision tree classification and regression tree (CART), random forest (RF), support vector machine (SVM), linear regression (LR), and super learner [SL; combined CART, RF and extreme gradient boosting (XGBoost)] for 28-day death in patients with septic shock was compared, and the best algorithm model was selected. The optimal predictive variables were determined by intersecting the results from LASSO regression, RF, and XGBoost algorithms, and a predictive model was constructed. The predictive efficacy of the model was validated by drawing receiver operator characteristic curve (ROC curve), the accuracy of the model was assessed using calibration curves, and the practicality of the model was verified through decision curve analysis (DCA). A total of 3 295 patients with septic shock were included, with 2 164 surviving and 1 131 dying within 28 days, resulting in a mortality of 34.32%. Of these, 2 307 were in the training set (with 792 deaths within 28 days, a mortality of 34.33%), and 988 in the validation set (with 339 deaths within 28 days, a mortality of 34.31%). Five machine learning models were established based on the training set data. After including variables at three aspects, the area under the ROC curve (AUC) of RF, SVM, and LR machine learning algorithm models for predicting 28-day death in septic shock patients in the validation set was 0.823 [95% confidence interval (95%CI) was 0.795-0.849], 0.823 (95%CI was 0.796-0.849), and 0.810 (95%CI was 0.782-0.838), respectively, which were higher than that of the CART algorithm model (AUC = 0.750, 95%CI was 0.717-0.782) and SL algorithm model (AUC = 0.756, 95%CI was 0.724-0.789). Thus above three algorithm models were determined to be the best algorithm models. After integrating variables from three aspects, 16 optimal predictive variables were identified through intersection by LASSO regression, RF, and XGBoost algorithms, including the highest pH value, the highest albumin (Alb), the highest body temperature, the lowest lactic acid (Lac), the highest Lac, the highest serum creatinine (SCr), the highest Ca2+, the lowest hemoglobin (Hb), the lowest white blood cell count (WBC), age, simplified acute physiology score III (SAPS III), the highest WBC, acute physiology score III (APS III), the lowest Na+, body mass index (BMI), and the shortest activated partial thromboplastin time (APTT) within 24 hours of ICU admission. ROC curve analysis showed that the Logistic regression model constructed with above 16 optimal predictive variables was the best predictive model, with an AUC of 0.806 (95%CI was 0.778-0.835) in the validation set. The calibration curve and DCA curve showed that this model had high accuracy and the highest net benefit could reach 0.3, which was significantly outperforming traditional models based on single functional score [APS III score, SAPS III score, and sequential organ failure assessment (SOFA) score] with AUC (95%CI) of 0.746 (0.715-0.778), 0.765 (0.734-0.796), and 0.625 (0.589-0.661), respectively. The Logistic regression model, constructed using 16 optimal predictive variables including pH value, Alb, body temperature, Lac, SCr, Ca2+, Hb, WBC, SAPS III score, APS III score, Na+, BMI, and APTT, is identified as the best predictive model for the 28-day death risk in patients with septic shock. Its performance is stable, with high discriminative ability and accuracy.

Cortisol-to-DHEAS awakening response ratio in people with dementia and family caregivers: Associations with age, dementia severity and agitation

CT-DERIVED BODY COMPOSITION AND MACHINE LEARNING FOR PROGNOSTIC IMPROVEMENT IN COLORECTAL CANCER

Summary A meticulous interpretation of steady-state or unsteady-state relative permeability (Kr) experimental data is required to determine a complete set of Kr curves. In this work, different machine learning (ML) models were developed to assist in a faster estimation of these curves from steady-state drainage coreflooding experimental runs. These ML algorithms include gradient boosting (GB), random forest (RF), extreme gradient boosting (XGB), and deep neural network (DNN) with a main focus on and comparison of the two latter algorithms (XGB and DNN). Based on existing mathematical models, a leading-edge framework was developed where a large database of Kr and capillary pressure (Pc) curves were generated. This database was used to perform thousands of coreflood simulation runs representing oil-water drainage steady-state experiments. The results obtained from these simulation runs, mainly pressure drop along with other conventional core analysis data, were used to estimate analytical Kr curves based on Darcy’s law. These analytically estimated Kr curves along with the previously generated Pc curves were fed as features into the ML model. The entire data set was split into 80% for training and 20% for testing. The k-fold cross-validation technique was applied to increase the model’s accuracy by splitting 80% of the training data into 10 folds. In this manner, for each of the 10 experiments, nine folds were used for training and the remaining fold was used for model validation. Once the model was trained and validated, it was subjected to blind testing on the remaining 20% of the data set. The ML model learns to capture fluid flow behavior inside the core from the training data set. In terms of applicability of these ML models, two sets of experimental data were needed as input; the first was the analytically estimated Kr curves from the steady-state drainage coreflooding experiments, while the other was the Pc curves estimated from centrifuge or mercury injection capillary pressure (MICP) measurements. The trained/tested model was then able to estimate Kr curves based on the experimental results fed as input. Furthermore, to test the performance of the ML model when only one set of experimental data is available to an end user, a recurrent neural network (RNN) algorithm was trained/tested to predict Kr curves in the absence of Pc curves as an input. The performance of the three developed models (XGB, DNN, and RNN) was assessed using the values of the coefficient of determination (R2) along with the loss calculated during training/validation of the model. The respective crossplots along with comparisons of ground truth vs. artificial intelligence (AI)-predicted curves indicated that the model is capable of making accurate predictions with an error percentage between 0.2% and 0.6% on history-matching experimental data for all three tested ML techniques. This implies that the AI-based model exhibits better efficiency and reliability in determining Kr curves when compared to conventional methods. The developed ML models by no means replace the need to conduct drainage coreflooding or centrifuge experiments but act as an alternative to existing commercial platforms that are used to interpret experimental data to predict Kr curves. The two main advantages of the developed ML models are their capability of predicting Kr curves within a matter of a few minutes as well as with limited intervention from the end user. The results also include a comparison between classical ML approaches, shallow neural networks, and DNNs in terms of accuracy in predicting the final Kr curves. The research presented here is an extension of the state-of-the-art framework proposed by Mathew et al. (2021). However, the two main aspects of the current study are the application of deep learning for the prediction of Kr curves and the application of feature engineering. The latter not only reduces the training/testing time for the ML models but also enables the end user to obtain the final predictions with the least set of experimental data. The various models discussed in this research work currently focus on the prediction of Kr curves for drainage steady-state experiments; however, the work can be extended to capture the imbibition cycle as well.

BackgroundSleep disturbances in Alzheimer’s disease impact people with dementia (PWD) and caregivers (CGS) who have primary responsibility for care provision. However, it is unknown which aspects of dementia including cognitive decline and behavioral disturbance contribute most to sleep disturbances in patients and caregivers. This pilot study aims to discover what contributes most to sleep deficits, and whether these factors are consistent across both PWD and CGS.MethodPatients: Six dyads with PWD and their primary CGS. Cognitive/behavioral variables: the Montreal Cognitive Assessment (MoCA) (PWD) and the Neuropsychiatric Inventory Questionnaire (NPI‐Q) completed by CGS. Polysomnography: recorded and scored for 1 night for each dyad member separately. Actigraphy: Wrist actigraphy recorded for 14 days in both members of the dyad divided into fixed rest (2200‐0600) and activity (0600‐2200) periods. Sleep variables: total sleep time (TST), sleep onset latency (SOL), nighttime activity as measured by actigraphy.ResultPreliminary results indicate that severity of cognitive dysfunction measured by the MoCA correlates with sleep time (CG r = 0.6224, PWD r = 0.5421). However, presence of neuropsychiatric symptoms and caregiver distress as measured by the NPIQ correlate with sleep onset latency (CG r = 0.9432*, PWD r = 0.7819) and nighttime activity in both CGs and PWD (CG r = 0.8467, PWD r = 0.8708). * = p<0.05.ConclusionThis pilot study of six PWD/CGS dyads examines the relationship between cognition, behavioral disturbance, CGS response to the behavioral disturbance, and sleep. This data indicates that sleep disturbances in PWD and their CGs cannot be explained exclusively by worsening cognitive function and likely have stronger dependence on the neuropsychiatric well‐being of the PWD and CGS emotional response. This dyadic model is unique in that it looks at the interdependence of the PWD/CGS dyad using both objective measures of sleep, cognition, and behavioral disturbance as well as subjective CGS stress. Data from additional dyads is currently being collected to increase the sample size along with the inclusion of additional variables in order to increase understanding of the relationship between PWD characteristics, caregiver burden and sleep disturbances.

A complex motor skill highly relevant to mobility in everyday life (e.g., sit-to-stand [STS] transfer) has not yet been addressed in studies on motor learning in people with dementia (PwD). To determine whether a dementia-specific motor learning exercise program enables PwD to learn compensatory STS maneuvers commonly taught in geriatric rehabilitation therapy to enhance patients' STS ability. Ninety-seven patients with mild-to-moderate dementia (Mini-Mental State Examination: 21.9±2.9 points) participated in a double-blinded, randomized, placebo-controlled trial with 10-week intervention and 3-month follow-up period. The intervention group (IG, n = 51) underwent a motor learning exercise program on compensatory STS maneuvers specifically designed for PwD. The control group (CG, n = 46) performed a low-intensity motor placebo activity. Primary outcomes were scores of the Assessment of Compensatory Sit-to-stand Maneuvers in People with Dementia (ACSID), which covers the number of recalled and initiated, and of effectively performed compensatory STS maneuvers. Secondary outcomes included temporal and kinematic STS characteristics measured by a body-fixed motion sensor (BFS, DynaPort® Hybrid). The IG significantly improved in all ACSID scores compared to the CG (p < 0.001). Secondary analysis confirmed learning effects for all BFS-based outcomes (p < 0.001-0.006). Learning gains were sustained during follow-up for most outcomes. People with mild-to-moderate dementia can learn and retain compensatory STS maneuvers in response to a dementia-specific motor learning exercise program. This is the first study that demonstrated preserved motor learning abilities in PwD by using a motor skill highly relevant to everyday life.

Preventing the expansion of perihematomal edema (PHE) represents a novel strategy for the improvement of neurological outcomes in intracerebral hemorrhage (ICH) patients. Our goal was to predict early and delayed PHE expansion using a machine learning approach. We enrolled 550 patients with spontaneous ICH to study early PHE expansion, and 389 patients to study delayed expansion. Two imaging researchers rated the shape and density of hematoma in non-contrast computed tomography (NCCT). We trained a radiological machine learning (ML) model, a radiomics ML model, and a combined ML model, using data from radiomics, traditional imaging, and clinical indicators. We then validated these models on an independent dataset by using a nested 4-fold cross-validation approach. We compared models with respect to their predictive performance, which was assessed using the receiver operating characteristic curve. For both early and delayed PHE expansion, the combined ML model was most predictive (early/delayed AUC values were 0.840/0.705), followed by the radiomics ML model (0.799/0.663), the radiological ML model (0.779/0.631), and the imaging readers (reader 1: 0.668/0.565, reader 2: 0.700/0.617). We validated a machine learning approach with high interpretability for the prediction of early and delayed PHE expansion. This new technique may assist clinical practice for the management of neurocritical patients with ICH. • This is the first study to use artificial intelligence technology for the prediction of perihematomal edema expansion. • A combined machine learning model, trained on data from radiomics, clinical indicators, and imaging features associated with hematoma expansion, outperformed all other methods.

Improving autocoding performance of rare categories in injury classification: Is more training data or filtering the solution?

General practitioners (GPs) play a key role in the care of people with dementia (PwD). However, the potentials of evidence-based dementia guidelines in primary care remain unclear. The main objective of the present study was to examine the role of guideline-based dementia care in general practices and to evaluate the effectiveness of a tablet-based intervention aiming at improving guideline-based care. A two arm, cluster-randomized controlled trial (cRCT) with an intervention group (tablet-based intervention) and a control group (treatment as usual) was conducted from April 15, 2018, to July 31, 2021, in Berlin and the surrounding area. Primary outcome was defined as adherence to dementia guideline recommendations after 9 months. Secondary outcomes included various health outcomes assessed in PwD (e.g., quality of life). Linear Mixed Models were used to analyse cross sectional baseline data. Primary and secondary outcomes were analysed by intention-to-treat analysis and using Mixed Models. Data of N = 28 GPs and N = 91 PwD was analysed. PwD were on average 80.5 years old (SD = 6.3), 59% were female. Self-reported overall guideline adherence of GPs was on average 71% (SD = 19.4, range: 25–100). Further, lower guideline adherence was found to be significantly associated with higher numbers of patients (γ10 = − 5.58, CI = − 10.97, − 0.19, p = .04). The effectiveness of the cRCT is still under evaluation. Final results will be presented and discussed at the GSA Annual Scientific Meeting. Further, implications and conclusions of the trial will be provided.