Predicting quantity of cannabis smoked in daily life: An exploratory study using machine learning.

BACKGROUND: Cannabis use is prevalent in the United States and is associated with a host of negative consequences. Importantly, a robust indicator of negative consequences is the amount of cannabis consumed. METHODS: Data were obtained from fifty-two adult, regular cannabis flower users (3+ times per week) recruited from the community; participants completed multiple ecological momentary assessment (EMA) surveys each day for 14 days. In this exploratory study, we used various machine learning algorithms to build models to predict the amount of cannabis smoked since participants' last report including forty-three EMA measures of mood, impulsivity, pain, alcohol use, cigarette use, craving, cannabis potency, cannabis use motivation, subjective effects of cannabis, social context, and location in daily life. RESULTS: Our best-fitting model (Gradient Boosted Trees; 71.15% accuracy, 72.46% precision) found that affects, subjective effects of cannabis, and cannabis use motives were among the best predictors of cannabis use amount in daily life. The social context of being with others, and particularly with a partner or friend, was moderately weighted in the final prediction model, but contextual items reflecting location were not strongly weighted in the final prediction model, the one exception being not at work. CONCLUSIONS: Machine learning approaches can help identify additional environmental and psychological phenomena that may be clinically-relevant to cannabis use.

Measuring affect in daily life: A multilevel psychometric evaluation of the PANAS-X across four ecological momentary assessment samples.

While there is strong evidence for the psychometric reliability of the Positive and Negative Affect Schedule-Expanded Form (PANAS-X) in cross-sectional studies, the between- and within-person psychometric performance of the PANAS-X in an intensive longitudinal framework is less understood. As affect is thought to be dynamic and responsive to context, this study investigated the multilevel reliability of PANAS-X Positive Affect, Negative Affect, Fear, Sadness, and Hostility scales. Generalizability theory and structural equation modeling techniques (coefficient ω) were employed in four ecological momentary assessment samples (N = 309; 41,261 reports). Results demonstrate that the PANAS-X scales, including short versions of the Positive and Negative Affect scales, can reliably detect between-person differences. PANAS-X scales also were able to reliably measure within-person change, though these estimates may be impacted by scale content and study design. These results support the use of the PANAS-X in daily life research to intensively measure affect in the natural environment. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

Agreement between momentary and retrospective reports of cannabis use and alcohol use: Comparison of ecological momentary assessment and timeline followback indices.

OBJECTIVE: This study compares three methods of cannabis and of alcohol use assessment in a sample of regular cannabis users: (a) ecological momentary assessment (EMA) repeated momentary surveys aggregated to the daily level, (b) EMA morning reports (MR) where participants reported on their total use from the previous day, and (c) retrospective timeline followback (TLFB) interviews covering the same period of time as the EMA portion of the study. We assessed the overall correspondence between these methods in terms of cannabis and alcohol use occasions and also investigated predictors of agreement between methods. METHOD: Forty-nine individuals aged 18-50 (Mage = 24.49, 49% female, 84% White) who reported regular cannabis use completed a 14-day EMA study. At the end of the EMA period, participants returned to the laboratory to complete a TLFB (administered via computer) corresponding to the same dates of the EMA period. RESULTS: Daily aggregated EMA and TLFB reports showed a low to modest agreement for both alcohol and cannabis use. Overall, agreement between EMA and MR was better than agreement between EMA and TLFB, likely because less retrospection is required when only reporting on behavior from the previous day. Quantity and frequency of use differentially predicted agreement across reporting methods when assessing alcohol compared to cannabis. When reporting cannabis use, but not alcohol use, individuals who used more demonstrated higher agreement between EMA and TLFB. CONCLUSIONS: Results suggest that retrospective reporting methods assessing alcohol and cannabis should not be considered a direct "substitute" for momentary or daily assessments. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

Using ecological momentary assessment and a portable device to quantify standard tetrahydrocannabinol units for cannabis flower smoking.

AIMS: To evaluate the feasibility and validity of a new method of quantifying cannabis flower use, integrating the amount of cannabis flower smoked, and the potency of the cannabis flower. DESIGN: Ecological momentary assessment (EMA) for 14 days. SETTING: Participants' daily lives in Columbia, Missouri, USA. PARTICIPANTS: A total of 50 community participants, who were regular cannabis flower smokers (48% female). MEASUREMENTS: Momentary subjective intoxication ratings following cannabis flower smoking; momentary quantity of cannabis flower smoked; potency of cannabis flower smoked in terms of percentage of tetrahydrocannabinol (THC) concentration assessed with a portable device, the Purpl Pro; and time since finished smoking. FINDINGS: Participants completed our field testing of their cannabis flower (96.2%) and were compliant with our 2-week EMA protocol (73% for random prompts and 91% for morning reports). Momentary subjective intoxication ratings trended down as a function of time since smoking (r = -0.10, P = 0.004, 95% CI, [-0.17, -0.03]). Multi-level model (MLM) results indicated the momentary standard THC units (mg THC) were positively associated with momentary subjective intoxication ratings (b = 0.01, P = 0.03, 95% CI, [0.01, 0.012]). CONCLUSIONS: There is evidence to support the feasibility and initial validity of a new method of quantifying cannabis flower use into standard tetrahydrocannabinol units. Researchers investigating the effects of cannabis flower use on a range of outcomes (e.g. neurobehavioral effects, emotional sequelae, and driving impairment) as well as in clinical treatment trials might adopt this method to provide estimates of cannabis flower use.

Relations between lab indices of emotion dysregulation and negative affect reactivity in daily life in two independent studies.

BACKGROUND: We investigated the extent to which physiological/biological measures of emotion dysregulation collected in the lab, resting respiratory sinus arrhythmia (RSA) in Study 1 and amygdala activation in response to negative stimuli in Study 2, combined with daily measures of interpersonal stressors predicted negative emotional states in outpatients better than the stressors alone. METHODS: Participants were adult outpatients with emotional distress disorders (N=30 individuals in Study 1, and N=26 women in Study 2). After completing a laboratory session that collected physiological/biological measures of emotion dysregulation, participants then completed 1-3 weeks of ambulatory assessment during which they reported on interpersonal stressors and negative affective states several times per day. RESULTS: Laboratory measures of emotion dysregulation were largely unrelated to either momentary or mean levels of daily-life hostility, sadness, and fear in both studies. However, resting RSA significantly moderated the association between day-level interpersonal stressors and momentary fear such that low resting RSA strengthened this association. Similarly, amygdala activation tended to moderate this relationship in the predicted direction. LIMITATIONS: Both samples were relatively small and focused on only a limited set of diagnoses associated with emotion dysregulation. Only two possible physiological/biological markers of emotion dysregulation were examined. CONCLUSIONS: The current studies support the collection of physiological/biological data on emotion dysregulation when indexing daily-life emotion dysregulation as the degree of emotional reactivity to stressors in daily life among outpatients with emotional distress disorders.

A daily-life study of interpersonal stressors and alcohol use in individuals with borderline personality disorder and community controls.

BACKGROUND: Interpersonal stressors (ISs) are major factors in relapse in alcohol use disorder (AUD) and are theorized to play a role in drinking behaviors. Past work has examined this association using ecological momentary assessment (EMA), but the unique effects of rejections and disagreements on alcohol use are unknown. Research suggests the two ISs functionally differ and may display distinct associations with drinking. Further, these associations may differ in people with borderline personality disorder (BPD), a population reporting frequent IS and co-occurring AUD. METHODS: 113 drinkers (community: n = 59; BPD: n = 54) reported alcohol use and ISs using EMA for 21 days. Using generalized estimating equations, we expected that rejection and disagreement would predict increased likelihood of drinking each day. We examined both cumulative (throughout each day) and immediate momentary effects of ISs predicting subsequent drinking on that same day. Further, we predicted that these associations would be stronger in individuals with BPD. RESULTS: Greater rejections throughout the day were associated with a reduced likelihood of drinking that day (OR = 0.56, 95 % CI:[0.32, 0.97], p < .040). In contrast, disagreements immediately prior to drinking were associated with an increased likelihood of drinking that day (OR = 0.60, 95 % CI:[1.02, 2.50], p = .039). However, the effect of disagreement on drinking was moderated by BPD diagnosis (OR = 2.56, 95 % CI:[1.13, 5.80], p = .025), such that the effect was only present for individuals with BPD. CONCLUSIONS: Assessing ISs as an aggregate predictor may mask potentially opposite effects on alcohol use. Additionally, disagreements may be a risk factor for subsequent alcohol use in BPD.

Considerations for the Ethical Implementation of Psychological Assessment Through Social Media via Machine Learning.

The ubiquity of social media usage has led to exciting new technologies such as machine learning. Machine learning is poised to change many fields of health, including psychology. The wealth of information provided by each social media user in combination with machine learning technologies may pave the way for automated psychological assessment and diagnosis. Assessment of individuals' social media profiles using machine learning technologies for diagnosis and screening confers many benefits (i.e., time and cost efficiency, reduced recall bias, information about an individual's emotions and functioning spanning months or years, etc.); however the implementation of these technologies will pose unique challenges to the professional ethics of psychology. Namely, psychologists must understand the impact of these assessment technologies on privacy and confidentiality, informed consent, recordkeeping, bases for assessments, and diversity and justice. This paper offers a brief review of the current applications of machine learning technologies in psychology and public health, provides an overview of potential implementations in clinical settings, and introduces ethical considerations for professional psychologists. This paper presents considerations which may aid in the extension of the current Ethical Principles of Psychologists and Code of Conduct to address these important technological advancements in the field of clinical psychology.