Comprehensive evaluation of machine learning algorithms for predicting sleep-wake conditions and differentiating between the wake conditions before and after sleep during pregnancy based on heart rate variability.

Introduction: Perinatal women tend to have difficulties with sleep along with autonomic characteristics. This study aimed to identify a machine learning algorithm capable of achieving high accuracy in predicting sleep-wake conditions and differentiating between the wake conditions before and after sleep during pregnancy based on heart rate variability (HRV). Methods: Nine HRV indicators (features) and sleep-wake conditions of 154 pregnant women were measured for 1 week, from the 23rd to the 32nd weeks of pregnancy. Ten machine learning and three deep learning methods were applied to predict three types of sleep-wake conditions (wake, shallow sleep, and deep sleep). In addition, the prediction of four conditions, in which the wake conditions before and after sleep were differentiated-shallow sleep, deep sleep, and the two types of wake conditions-was also tested. Results and Discussion: In the test for predicting three types of sleep-wake conditions, most of the algorithms, except for Naïve Bayes, showed higher areas under the curve (AUCs; 0.82-0.88) and accuracy (0.78-0.81). The test using four types of sleep-wake conditions with differentiation between the wake conditions before and after sleep also resulted in successful prediction by the gated recurrent unit with the highest AUC (0.86) and accuracy (0.79). Among the nine features, seven made major contributions to predicting sleep-wake conditions. Among the seven features, "the number of interval differences of successive RR intervals greater than 50 ms (NN50)" and "the proportion dividing NN50 by the total number of RR intervals (pNN50)" were useful to predict sleep-wake conditions unique to pregnancy. These findings suggest alterations in the vagal tone system specific to pregnancy.

The return of individual genomic results to research participants: design and pilot study of Tohoku Medical Megabank Project.

Certain large genome cohort studies attempt to return the individual genomic results to the participants; however, the implementation process and psychosocial impacts remain largely unknown. The Tohoku Medical Megabank Project has conducted large genome cohort studies of general residents. To implement the disclosure of individual genomic results, we extracted the potential challenges and obstacles. Major challenges include the determination of genes/disorders based on the current medical system in Japan, the storage of results, prevention of misunderstanding, and collaboration of medical professionals. To overcome these challenges, we plan to conduct multilayer pilot studies, which deal with different disorders/genes. We finally chose familial hypercholesterolemia (FH) as a target disease for the first pilot study. Of the 665 eligible candidates, 33.5% were interested in the pilot study and provided consent after an educational "genetics workshop" on the basic genetics and medical facts of FH. The genetics professionals disclosed the results to the participants. All positive participants were referred to medical care, and a serial questionnaire revealed no significant psychosocial distress after the disclosure. Return of genomic results to research participants was implemented using a well-prepared protocol. To further elucidate the impact of different disorders, we will perform multilayer pilot studies with different disorders, including actionable pharmacogenomics and hereditary tumor syndromes.

Heart Rate Information-Based Machine Learning Prediction of Emotions Among Pregnant Women.

In this study, the extent to which different emotions of pregnant women can be predicted based on heart rate-relevant information as indicators of autonomic nervous system functioning was explored using various machine learning algorithms. Nine heart rate-relevant autonomic system indicators, including the coefficient of variation R-R interval (CVRR), standard deviation of all NN intervals (SDNN), and square root of the mean squared differences of successive NN intervals (RMSSD), were measured using a heart rate monitor (MyBeat) and four different emotions including "happy," as a positive emotion and "anxiety," "sad," "frustrated," as negative emotions were self-recorded on a smartphone application, during 1 week starting from 23rd to 32nd weeks of pregnancy from 85 pregnant women. The k-nearest neighbor (k-NN), support vector machine (SVM), logistic regression (LR), random forest (RF), naïve bayes (NB), decision tree (DT), gradient boosting trees (GBT), stochastic gradient descent (SGD), extreme gradient boosting (XGBoost), and artificial neural network (ANN) machine learning methods were applied to predict the four different emotions based on the heart rate-relevant information. To predict four different emotions, RF also showed a modest area under the receiver operating characteristic curve (AUC-ROC) of 0.70. CVRR, RMSSD, SDNN, high frequency (HF), and low frequency (LF) mostly contributed to the predictions. GBT displayed the second highest AUC (0.69). Comprehensive analyses revealed the benefits of the prediction accuracy of the RF and GBT methods and were beneficial to establish models to predict emotions based on autonomic nervous system indicators. The results implicated SDNN, RMSSD, CVRR, LF, and HF as important parameters for the predictions.

Maternity Log study: a longitudinal lifelog monitoring and multiomics analysis for the early prediction of complicated pregnancy.

PURPOSE: A prospective cohort study for pregnant women, the Maternity Log study, was designed to construct a time-course high-resolution reference catalogue of bioinformatic data in pregnancy and explore the associations between genomic and environmental factors and the onset of pregnancy complications, such as hypertensive disorders of pregnancy, gestational diabetes mellitus and preterm labour, using continuous lifestyle monitoring combined with multiomics data on the genome, transcriptome, proteome, metabolome and microbiome. PARTICIPANTS: Pregnant women were recruited at the timing of first routine antenatal visits at Tohoku University Hospital, Sendai, Japan, between September 2015 and November 2016. Of the eligible women who were invited, 65.4% agreed to participate, and a total of 302 women were enrolled. The inclusion criteria were age ≥20 years and the ability to access the internet using a smartphone in the Japanese language. FINDINGS TO DATE: Study participants uploaded daily general health information including quality of sleep, condition of bowel movements and the presence of nausea, pain and uterine contractions. Participants also collected physiological data, such as body weight, blood pressure, heart rate and body temperature, using multiple home healthcare devices. The mean upload rate for each lifelog item was ranging from 67.4% (fetal movement) to 85.3% (physical activity), and the total number of data points was over 6 million. Biospecimens, including maternal plasma, serum, urine, saliva, dental plaque and cord blood, were collected for multiomics analysis. FUTURE PLANS: Lifelog and multiomics data will be used to construct a time-course high-resolution reference catalogue of pregnancy. The reference catalogue will allow us to discover relationships among multidimensional phenotypes and novel risk markers in pregnancy for the future personalised early prediction of pregnancy complications.

Integrated Database And Knowledge Base For Genomic Prospective Cohort Study In Tohoku Medical Megabank Toward Personalized Prevention And Medicine.

The Tohoku Medical Megabank project is a national project to revitalization of the disaster area in the Tohoku region by the Great East Japan Earthquake, and have conducted large-scale prospective genome-cohort study. Along with prospective genome-cohort study, we have developed integrated database and knowledge base which will be key database for realizing personalized prevention and medicine.

Development of clinical ontology for mood disorder with combination of psychomedical information.

We have developed a new educational/clinical ontology named the "Haghighi-Koeda Mood Disorder Ontology", which involves both medical and psychological approaches for mood disorders in order to promote the exchange of information between psychiatrists and psychologists. Data was gathered from more than 5000 articles published in journals and websites specialized in life science. We evaluated and selected articles which were related to 4 main categories of mood disorders. Using Protege 3.4 beta, information related to mood disorders was classified by class/subclass tree in an ontological structure. Then we developed a web-based interface system on the internet enabling the implementation of the ontology. In addition, we have designed an online scale for automated diagnosis of mood disorder. For evaluating experiments, we compare this ontology with "Decisionbase" of which content deals with mood disorders. Evaluation was in accordance with our selected criteria via the AHP (Analysis of Hierarchical Processing) method. The results demonstrated the noteworthy superiority of our ontology. We believe that combining knowledge of medical science with that in psychological fields is a key to improving the quality of diagnosis and promoting appropriate treaTMent in all psychiatric disorders.