Speak and you shall predict: speech at initial cocaine abstinence as a biomarker of long-term drug use behavior

Abstract

Importance: Valid biomarkers that can predict longitudinal clinical outcomes at low cost are a holy grail in psychiatric research, promising to ultimately be used to optimize and tailor intervention and prevention efforts.

Objective: To determine if baseline linguistic markers in natural speech, as compared to non-speech clinical and demographic measures, can predict drug use severity measures at future sessions in initially abstinent individuals with cocaine use disorder (iCUD).

Design: A longitudinal cohort study (August 2017 - March 2020), where baseline measures were used to predict outcomes collected at three-month intervals for up to one year of follow-up.

Participants: Eighty-eight initially abstinent iCUD were studied at baseline; 57 (46 male, age 50.7+/-7.9 years) came back for at least another session.

Main outcomes and measures: Outcomes were self-reported symptoms of withdrawal, craving, abstinence duration and frequency of cocaine use in the past 90 days at each study session. The predictors were derived from 5-min recordings of vocal descriptions of the positive consequences of abstinence and the negative consequences of using cocaine; the baseline cocaine and other common drug use measures, demographic and neuropsychological variables were used for comparison.

Results: Models using the non-speech variables showed the best predictive performance at three(r>0.45, P<2×10^-3) and six months follow-up (r>0.37, P<3×10^-2). At 12 months, the natural language processing-based model showed significant correlations with withdrawal (r=0.43, P=3×10^-2), craving (r=0.72, P=5×10^-5), days of abstinence (r=0.76, P=1×10^-5), and cocaine use in the past 90 days (r=0.61, P=2×10^-3), significantly outperforming the other models for abstinence prediction.

Conclusions and relevance: At short time intervals, maximal predictive power was obtained with models that used baseline drug use (in addition to demographic and neuropsychological) measures, potentially reflecting a slow rate of change in these measures, which could be estimated by linear functions. In contrast, short speech samples predicted longer-term changes in drug use, implying deeper penetrance by potentially capturing non-linear dynamics over longer intervals. Results suggest that, compared to the common outcome measures used in clinical trials, speech-based measures could be leveraged as better predictors of longitudinal drug use outcomes in initially abstinent iCUD, as potentially generalizable to other substance use disorders and related comorbidity.

Figure 1.. Approach used in this study to analyze recordings of individuals with cocaine use disorders to prospectively predict cocaine-related outcomes at 3, 6, 9, and 12 months follow-up.

Under predictive models, model 1 also includes drug use variables listed in the cocaine, alcohol and cigarette use section in Table 1; model 2 in addition includes all dependent measures obtained at baseline (including the four last variables in Table 1). STAI refers to the State-Trait Anxiety Inventory.

Figure 2:. Prediction of longitudinal outcomes in individuals with cocaine use disorder.

Tables at the left show the performance of the models that include baseline demographic, neuropsychological and drug use measures (non-NLP Model 1), in addition to the dependent/to be predicted measures obtained at baseline (non-NLP Model 2) and NLP models for positive consequences of abstinence (PC) and negative consequences of drug use (NC) for predicting four selected drug use measures (withdrawal, craving, abstinence days, and cocaine use in the last 90 days) at subsequent visits: 3 (N=50), 6 (N=36), 9 (N=35) and 12 (N=25) months after baseline. Symbols next to the values indicate that the model is statistically significant with regard to the null hypothesis (*) and when comparing between non-NLP and NLP models (^†). Plots on the right show a different representation of the tables (non-NLP Model 1 = an orange dashed line; non-NLP Model 2= a green dotted line; NLP models (the best of the NC or PC prompts) = a blue solid line).

Figure 3:. Predictive features of the first visit interview for: (a) withdrawal in the fifth visit (12 months), (b) craving in the fifth visit (12 months).

Larger positive weights predict more withdrawal symptoms and craving. Both models are based on the PC section of the interviews.

Figure 4:. Predictive features of the first visit interview for: (a) days of abstinence in the fifth visit (12 months), (b) the previous 90 days of cocaine use in the fifth visit (12 months).

Larger positive weights predict days of abstinence and more days of cocaine use. Both models are based on the NC section of the interviews.