Purpose
The aim of this research was to examine heterogeneity of Health-related Quality of Life (HrQOL) in Cancer Survivors (both undergoing and completed treatment) using latent profile analysis and to determine whether these groups differed by demographic and health characteristics.

Methods
Participants(n=229) recruited through an oncology day ward and outpatient department in a local hospital, completed height, weight and handgrip measures as well as the validated patient generated subjective global assessment and EORTC-QLQ-C30 questionnaires. A latent profile analysis was performed to identify subgroups based on HrQOL domain scores. Multinominal Logistic Regression was conducted to determine the relationship between these subgroups and demographic and health characteristics.

Results
Three latent subtypes were identified: (1)high quality of life(n=122, 52.8%); (2)compromised quality of life(n=79, 34.2%) and (3)low quality of life(n=30, 12.99%). All subtypes scored lower for functioning scales (with the exception of the higher quality of life group for physical, role and emotional functioning) and higher for symptom scales then the reference norm population. There were large clinically meaningful differences between the high quality of life group and the low quality of life group for all HrQOL scales.

Those in the low quality of life group were slightly younger than those in the high quality of life group(OR = 0.956, p < .05, CI = 0.917– 0.998). Workers were >7 times more likely to be in low quality of life than the high quality of life group. Compared to the high quality of life group, the odds of belonging to the compromised quality of life group decreased significantly by having higher handgrip strength (OR = .955, p < .05, CI = .924 - .988). The odds of belonging to the low quality of life group increased significantly for those with higher number of nutrition impact symptoms (NIS) (OR = 1.375, p < .05, CI = 1.004 – 1.883).

Conclusions
This is the first study to examine heterogeneity of HrQOL using latent profile analysis in Irish Cancer Survivors. In clinical practice understanding how aspects of HrQOL group together may allow clinicians to better understand and treat cancer survivors, informing more individualised nutrition care.

Purpose: Those with a cancer diagnosis report experiencing a wide range of nutrition impact symptoms with prevalence varying by study, group and cancer type. We aimed to identify groups of cancer survivors with specific patterns of nutrition impact symptoms.
Methods: 229 individuals attending oncology day ward and outpatient clinics completed a series of questionnaires and physical measurements. A latent class analysis was performed to identify subgroups based on 13 nutrition impact symptoms taken from the Patient Generated Subjective Global Assessment Short Form. The identified classes were subsequently compared using analysis of variance and chi-square tests, by sociodemographic, clinical and nutritional variables as well as by Global health status (GHS) and five functioning scales determined using the European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire (EORTC QLQ-C30).
Results: Three latent subtypes were identified: (1) Fatigue (n=58, 28%); (2) Low Symptom Burden (n=146, 64%) and (3) High Symptom Burden (n=25, 11%). Those in the High Symptom Burden group were more likely to be female, currently receiving any form of treatment and have consumed less food than usual in the last month compared to those in the Low Symptom Burden group. Those in the Fatigue group were more likely were more likely to have reported consuming less food in the previous month and less likely to have reported their food intake to be unchanged than those in the Low Symptom Burden group. Those who received their diagnosis two years+ ago were most likely to be classed in the Fatigue group. The EORTC-QLQ-C30 functioning and GHS scores were all significantly different between the three nutrition impact symptoms classes (p<0.001)
Conclusion: This is the first study to examine heterogeneity of nutrition impact symptoms in Irish Cancer Survivors. The findings of this work will inform and allow for more individualised nutrition care.

Purpose: Physical activity (PA) is associated with a reduction in mortality and recurrence risks in patients with cancer. Although there is increasing evidence that PA benefits the health of cancer patients before, during, and after treatment, the inclusion of PA recommendations is not currently a regular component of standard care. Therefore, the current study seeks to identify perceived knowledge, barriers, and facilitators of oncology care providers’ prescription of PA among cancer patients as described by the 5A (Assess, Advise, Agree, Assist, and Arrange) framework.

Methods: A qualitative research design with a phenomenological approach was used. A purposive sample of eleven oncology care providers representing diverse specialties (4 medical oncologists, 3 surgical oncologists, 3 oncology nurses, and 1 oncology nutritionist) was conducted via Zoom or phone calls. Participants were involved in cancer care across Nebraska, a midwestern region in the United States. Data were audio recorded, transcribed verbatim, and imported into qualitative software NVivo version 12. The responses were mapped using the 5A framework, where theory-driven deductive content analysis was used for data analysis.

Results/findings: The data revealed several themes and subthemes on the perceived knowledge, barriers, facilitators, and recommendations of PA among oncology care providers. The most identified PA barriers include a lack of awareness of PA promotion (limited knowledge about PA guidelines, limited training, and less priority to PA promotion); cancer-related symptoms (e.g., fatigue and pain); and logistic barriers (lack of community resources, time constraints, and cost). Further, the reported facilitators mentioned by oncology care providers were perceived health benefits (improve physical, mental, sleep, and survival and reduce recurrence), interdisciplinary support, and available resources in the community. Recommendations included education or training related to PA promotion, resources for patient education, and PA experts in the clinical team.

Conclusions: Oncology care providers acknowledged various barriers and facilitators highlighting the complex character of influencing advice and counseling on PA promotion. Having a thorough grasp of this will help healthcare providers and researchers identify specific characteristics that can be modified and targeted in future interventions. This will help improve the success of programs that attempt to promote PA in various settings.

Purpose: Consumer wearables often incorporate accelerometry to assess movement and photoplethysmography to capture heart rate (HR), which are both important markers of physical activity energy expenditure (PAEE). Given the widespread adoption of consumer wearables offering accessible high-granular accelerometry and HR data opens the possibility for developing a large-scale, reproducible, and device agnostic method for estimating PAEE. This study examined the performance of PAEE estimates based on the raw accelerometry and HR from research grade and consumer wearables compared to indirect calorimetry.
Methods: One hundred and ninety-six children (5-12yrs, 57% male, 71% White) participated in a 60-minute protocol consisting of 5-minute activities completed at varying intensities (e.g., seated, watching a video, walking, and playing tag). Children wore two of three consumer wearables (Apple Watch Series 7, Garmin Vivoactive 4S, Fitbit Sense) and a research-grade accelerometer (ActiGraph GT9X) on their non-dominant wrist, and a chest-placed, research-grade HR monitor (Actiheart 5, ECG), concurrently. Children also wore a COSMED K5 as a criterion measure of PAEE (i.e., V02/kg in ml/minute). Cross-sectional time series regression models with random intercepts were used to estimate minute-by-minute PAEE from features extracted from raw accelerometry and HR data. Child age, sex, weight, and height were also included in the models. R2 for the cross-sectional time series regression models, mean absolute error (MAE), and Lin’s Concordance Correlation Coefficient (CCC) were calculated to assess agreement between indirect calorimetry, consumer wearables, and research-grade indicators of PAEE.
Results/findings: For the research grade devices (i.e., ActiGraph accelerometry combined with Actiheart HR) MAE values were 4.7 (95CI=4.6, 4.9), while CCC was 0.83, and R2 was 0.78. Apple PAEE estimates had a MAE of 4.2 (95CI=4.1,4.3), CCC of 0.87, and R2 of 0.86. Garmin MAE was 4.4 (95CI=4.3,4.5), CCC of 0.85, and R2 of 0.82 Finally, Fitbit produced a MAE of 4.2 (95CI=4.1, 4.4), CCC of 0.84, and R2 of 0.80.
Conclusions: The raw accelerometry and HR data collected from consumer wearable devices predicted PAEE comparably to research grade accelerometry and HR. These outcomes support the possibility of deploying a device-agnostic approach to PAEE estimation using consumer grade sensor data in children.

Purpose

Understanding and optimizing compliance is critical in intensive longitudinal designs using ecological momentary assessment (EMA) where systematically missed prompts can severely threaten validity. This study leveraged hourly EMA data collected through multi-day bursts spaced every two weeks across 12 months from young adults in the U.S. This presentation will aim to: 1) elucidate time-variant contextual and behavioral predictors of EMA prompt compliance to identify refinements for future protocols and help interpret results; and 2) model decay in EMA prompt compliance across the year-long study to inform effective engagement and retention strategies for ambitious intensive longitudinal research crucial for advancing micro-temporal and idiographic analyses of behavior.

Methods

Young adults (n=246; ages 18-29; 55.1% female; 30.3% Hispanic) completed EMA measurement bursts every two weeks over 12 months. Each burst spanned four days with EMA surveys prompted approximately once per hour during wake time. At the same time, they participated in activity monitoring via smartwatch and completed end-of-day diaries. EMA surveys assessed health behaviors, contextual factors (e.g., location, social context), and psychological states. Prompt compliance, the primary outcome, was defined as responding within 10 minutes of the initial prompt. Multilevel logistic regression identified predictors of prompt compliance, including time-variant contextual factors (e.g., season, phone screen status, location, travel status) and behavioral factors (e.g., routine, prior 10-minute physical activity, affect, stress).

Results

Prompt compliance declined over time within participants. Prompts occurring while the phone screen was on (vs. off) were more likely to be completed. Compliance was lower during spring, fall, and winter (vs. summer); when at work/school (vs. home); when the individual was more physically active in the hour before prompt; and when the individual was not following their regular daily routine. Stress and positive affect did not significantly alter the odds of compliance in this study.

Conclusions

Results elucidate participation factors for interpreting findings and guiding effective retention strategies in year-long EMA research. These results also help to inform what situations may be underrepresented in the data. Adjustments to statistical analyses can be made and data imputations can help address these external validity issues.

Lingyi Fu (Presenting), Shandian Zhe (Corresponding), Yang Bai (Corresponding)

Background: While established connections exist between health-related behaviors and mental health, there is a notable gap in the literature regarding comprehensive examinations of the importance of multiple health-related behaviors associated with mental health among college students longitudinally.
Purpose: This study aims to examine the essential health-related behaviors (e.g., social interaction, diet, sleep, exercise, and substance use), both past and present, impacting overall mental health (i.e., well-being, mood, anxiety, and focus).
Methods: The participants reported their behaviors and mental health on a 16-item daily survey through a smartphone app. Ecological momentary assessment data were collected from 205 college students over 215 days from October 2020 to May 2021, resulting in 32,997 person-day observations. Machine learning approaches (i.e., Random forest, Gradient boosting, AdaBoost, Support vector machines, and KNeighbors) with SHAP (SHapley Additive exPlanations) value-based feature importance were utilized to examine and rank the importance of past and present behaviors on mental health.
Results: Most participants were freshmen (89.1%), female (74.3%), and Caucasian (90.3%). Random forest (F1 = 0.539—0.936) and Gradient boosting (F1 = 0.822—0.928) outperformed other approaches in predicting mental health. The top eight critical behaviors of each mental health were identified and compared. Quality of social interactions emerged as a paramount determinant of overall mental health (SAHP = 0.029—0.163), particularly influencing well-being and mood. The importance of sleep and meal quality (SAHP = 0.011—0.112) was underscored in shaping overall mental health than quantity (SAHP = 0.006—0.047). Moreover, the current engagement in gratitude moments held more significant sway over overall mental health (SAHP = 0.010—0.056), while previous engagement only held importance to anxiety (SAHP = 0.059) and focus (SHAP = 0.064). Water consumption, exercise participation, and screen time emerged on the important list for some mental health outcomes but not all (SAHP = 0.005—0.044). Furthermore, the frequency of marijuana and liquor used in previous days emerged as a noteworthy factor affecting well-being (SAHP = 0.005) and focus status (SAHP = 0.042), respectively.
Conclusion: This study provides a comprehensive understanding of the critical previous and current health-related behaviors associated with mental health.