Personalized Physician-Assisted Sleep Advice for Shift Workers: Machine Learning Approach (Preprint)

Опубликована: Авг. 2, 2024

In the modern economy, shift work is prevalent in numerous occupations. However, often conflicts with workers’ circadian rhythm and can result sleep disorder (SWSD). Proper management of SWSD emphasizes comprehensive patient-specific strategies some these are analogous to cognitive behavioral treatment insomnia (CBTI). this paper, we aim develop evaluate machine learning algorithms that predict physicians’ advice using wearable survey data. We developed an online system conveniently frequently provide individualized behavior CBTI elements for workers. Data were collected a period 5 weeks from workers ICU at two hospitals (N = 61) Japan. The data composed three modalities, (1) Fitbit data, (2) (3) advice. handcrafted physiological features raw identified clusters participants similar characteristics hierarchical clustering. After first week enrollment, physicians reviewed list 23 messages. implemented random forest (RF) models 7 most frequent messages given by physicians. tested our predictions under participant dependent independent settings analyzed important prediction. found distinguished shifts patterns. For clusters, having on day contributed low wellbeing scores day. Another cluster had days duration lowest quality when there was before midnight current Our prediction achieved higher F1 27 28 t-tests conducted, performance differences statistically significant P < .001 24 tests .05 3 compared baseline. analysis feature importance showed matched message sent participants. instance, (darken bedroom you go bed), primarily examined average brightness environment make predictions. Although requires physician input, accurate algorithm would be promising automating without hurting trustworthiness selected recommendations. limited popular ones among choices due rare occurrences remaining options. Therefore, further studies necessary gather enough enable less labels. UMIN Clinical Trials Registry UMIN000036122 (phase 1), UMIN000040547 2); https://tinyurl.com/dkfmmmje, https://upload.umin.ac.jp/cgi-open-bin/ctr_e/ctr_view.cgi?recptno=R000046284.

Язык: Английский

---

Personalized Physician-Assisted Sleep Advice for Shift Workers: Machine Learning Approach (Preprint)

JMIR Formative Research, Год журнала: 2024, Номер unknown

Опубликована: Авг. 2, 2024

In the modern economy, shift work is prevalent in numerous occupations. However, it often disrupts workers' circadian rhythms and can result sleep disorder. Proper management of disorder involves comprehensive patient-specific strategies, some which are similar to cognitive behavioral therapy for insomnia. Our goal was develop evaluate machine learning algorithms that predict physicians' advice using wearable survey data. We developed a web- app-based system provide individualized behavior based on insomnia workers. Data were collected 5 weeks from workers (N=61) intensive care unit at 2 hospitals Japan. The data comprised 3 modalities: Fitbit data, advice. After first week enrollment, physicians reviewed selected 1 messages list 23 options. handcrafted physiological features raw identified clusters participants with characteristics hierarchical clustering. explored models (random forest, light gradient-boosting machine, CatBoost) data-balancing approaches (no balancing, random oversampling, synthetic minority oversampling technique) selections 7 most frequent related bedroom brightness, smartphone use, nap duration. tested our predictions under participant-dependent participant-independent settings analyzed important prediction permutation importance Shapley additive explanations. found distinguished by shifts patterns. For example, one cluster had days low duration lowest quality when there day before midnight current day. achieved higher area precision-recall curve than baseline all settings. performance differences statistically significant (P<.001 13 tests P=.003 test). Sensitivity ranged 0.50 1.00, specificity varied between 0.44 0.93 across dataset split Feature analysis several matched corresponding sent. instance, message (darken you go bed), primarily examined average brightness environment make predictions. Although requires physician input, an accurate algorithm shows promise automatic without compromising trustworthiness recommendations. Despite its decent performance, currently limited popular messages. Further studies needed enable less labels.

Язык: Английский

---

Bevorzugen Frauen Face-to-Face-Beratung bei Insomnie?

Somnologie - Schlafforschung und Schlafmedizin, Год журнала: 2021, Номер 25(2), С. 151 - 154

Опубликована: Март 2, 2021

---

Psychosocial Features of Shift Work Disorder

Brain Sciences, Год журнала: 2021, Номер 11(7), С. 928 - 928

Опубликована: Июль 14, 2021

To better understand Shift Work Disorder (SWD), this study investigates insomnia, sleepiness, and psychosocial features of night workers. The compares workers with or without SWD to day insomnia. Seventy-nine 40 underwent diagnostic interviews for sleep disorders psychopathologies. They completed questionnaires a diary 14 days. design was observatory upon two factors: schedule (night, work) (good sleep, SWD/insomnia). Two-way ANCOVAs were conducted on variables, effect size calculated. clinical approach chosen led distinct groups Night slept several periods (main period after work, naps, nights days off). High total wake time low characterized in SWD. Most still complained sleepiness main sleep. Cognitive activation distinguished All other differences variables between similar to, but smaller than, the ones evaluation should consider all particular attention self-reported time, state level cognitive activation.

Язык: Английский

---

Anonymous Online Cognitive Behavioral Therapy for Sleep Disorders in Shift Workers – A Study Protocol for a Randomized Controlled Trial

Research Square (Research Square), Год журнала: 2021, Номер unknown

Опубликована: Июнь 29, 2021

Abstract Background: Many shift workers suffer from sleep issues, which negatively affect quality of life and performance. Scientifically evaluated, structured programs for prevention treatment are scarce. We developed an anonymous online cognitive behavioral therapy insomnia (CBT-I) program. After successful completion a feasibility study, we now start this prospective, randomized, controlled superiority trial to compare outcomes two parallel groups, namely intervention group waiting-list control-group. Additionally, will these those face-to-face CBT-I outpatient sample. Methods: Collaborating companies offer our their shift-working employees. Company physicians counseling services screen interested inclusion exclusion criteria. Participants receive access service, where they complete psychometric assessment random assignment either the or control group. providers be aware assignment. aim allocate at least N = 60 participants trial. The consists psychoeducation, restriction, stimulus control, relaxation techniques, individual feedback delivered via four e-mail contacts. During intervention, as well during waiting period, fill out weekly diaries. Immediately after program, post-intervention takes place. in able participate program all study assessments. To recruit additional sample, collaborating clinics provide six sessions standard ad-hoc sample working patients. expect both interventions have beneficial effects compared on following primary outcomes: self-reported symptoms depression insomnia, quality, daytime sleepiness. Conclusions: allows follow independently schedule location. Forthcoming results might contribute further improvement issues workers.

Язык: Английский

---