{"doi":"10.1101/2023.07.18.549548","title":"Speak and you shall predict: speech at initial cocaine abstinence as a biomarker of long-term drug use behavior","abstract":"Abstract\n \n Importance\n 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.\n \n \n Objective\n 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).\n \n \n Design\n 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.\n \n \n Participants\n 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.\n \n \n Main Outcomes and Measures\n 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.\n \n \n Results\n \n Models using the non-speech variables showed the best predictive performance at three(r>0.45,\n P\n <2×10\n -3\n ) and six months follow-up (r>0.37,\n P\n <3×10\n -2\n ). At 12 months, the natural language processing-based model showed significant correlations with withdrawal (r=0.43,\n P\n =3×10\n -2\n ), craving (r=0.72,\n P\n =5×10\n -5\n ), days of abstinence (r=0.76,\n P\n =1×10\n -5\n ), and cocaine use in the past 90 days (r=0.61,\n P\n =2×10\n -3\n ), significantly outperforming the other models for abstinence prediction.\n \n \n \n Conclusions and Relevance\n 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.\n \n \n Key Points\n \n Question\n Can natural language processing be leveraged to predict longitudinal drug use outcomes in individuals with substance use disorder?\n \n \n Findings\n In this prospective study that included initially abstinent individuals with cocaine use disorder (iCUD), models using demographics, neuropsychological measures and drug use patterns at baseline were compared to those using minimally structured short natural speech samples relating the positive consequences of abstinence and the negative consequences of using drugs, showing a differential prediction of outcomes measured up to one year later. At three and six months, the former outperformed speech models, including approximately 50% of the variability in craving and 40% in abstinence duration. At 12 months from baseline, speech models were superior, predicting 50% of the variability in abstinence duration.\n \n \n Meaning\n Speech variables derived through natural language processing can predict clinically meaningful drug use outcome measures in addiction, with greater value at longer intervals. The applicability of language modeling to aid in assessing treatment response and risk in drug addiction warrants further investigation in clinical settings.\n \n ","journal":null,"year":null,"id":626654,"datarank":0.20794415416798362,"base_score":1.3862943611198906,"endowment":1.3862943611198906,"self_citation_contribution":0.20794415416798362,"citation_network_contribution":0.0,"self_endowment_contribution":0.20794415416798362,"citer_contribution":0.0,"corpus_percentile":null,"corpus_rank":null,"citation_count":3,"citer_count":0,"citers_with_citation_signal":0,"citers_with_endowment":0,"datacite_reuse_total":0,"is_dataset":false,"is_dataset_confidence":null,"is_data_producer":false,"deposit_databanks":null,"is_oa":false,"file_count":0,"downloads":0,"has_version_chain":false,"published_date":null,"fair_score":null,"fair_percentile":null,"algorithm_id":"datarank_citation_only_1hop_v6","ranking_scope":"data_only","authors":[{"id":249475,"name":"Guillermo Cecchi","orcid":"0000-0003-1013-8348","position":1,"is_corresponding":false},{"id":692482,"name":"Sarah King","orcid":"0000-0001-6821-9234","position":2,"is_corresponding":false},{"id":1621078,"name":"Elif K. Eyigoz","orcid":null,"position":3,"is_corresponding":false},{"id":399165,"name":"Muhammad A. Parvaz","orcid":"0000-0002-2671-2327","position":4,"is_corresponding":false},{"id":1621081,"name":"Nelly Alia-Klein","orcid":null,"position":5,"is_corresponding":false},{"id":399167,"name":"Rita Z. Goldstein","orcid":"0000-0002-1680-229X","position":6,"is_corresponding":false},{"id":653148,"name":"Carla Agurto","orcid":"0000-0002-0617-4488","position":0,"is_corresponding":false}],"reference_count":0,"raw_metadata":{"has_enrichment":true,"resolved":true,"title":"Speak and you shall predict: speech at initial cocaine abstinence as a biomarker of long-term drug use behavior","abstract":"Abstract\n \n Importance\n 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.\n \n \n Objective\n 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).\n \n \n Design\n 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.\n \n \n Participants\n 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.\n \n \n Main Outcomes and Measures\n 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.\n \n \n Results\n \n Models using the non-speech variables showed the best predictive performance at three(r>0.45,\n P\n <2×10\n -3\n ) and six months follow-up (r>0.37,\n P\n <3×10\n -2\n ). At 12 months, the natural language processing-based model showed significant correlations with withdrawal (r=0.43,\n P\n =3×10\n -2\n ), craving (r=0.72,\n P\n =5×10\n -5\n ), days of abstinence (r=0.76,\n P\n =1×10\n -5\n ), and cocaine use in the past 90 days (r=0.61,\n P\n =2×10\n -3\n ), significantly outperforming the other models for abstinence prediction.\n \n \n \n Conclusions and Relevance\n 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.\n \n \n Key Points\n \n Question\n Can natural language processing be leveraged to predict longitudinal drug use outcomes in individuals with substance use disorder?\n \n \n Findings\n In this prospective study that included initially abstinent individuals with cocaine use disorder (iCUD), models using demographics, neuropsychological measures and drug use patterns at baseline were compared to those using minimally structured short natural speech samples relating the positive consequences of abstinence and the negative consequences of using drugs, showing a differential prediction of outcomes measured up to one year later. At three and six months, the former outperformed speech models, including approximately 50% of the variability in craving and 40% in abstinence duration. At 12 months from baseline, speech models were superior, predicting 50% of the variability in abstinence duration.\n \n \n Meaning\n Speech variables derived through natural language processing can predict clinically meaningful drug use outcome measures in addiction, with greater value at longer intervals. The applicability of language modeling to aid in assessing treatment response and risk in drug addiction warrants further investigation in clinical settings.\n \n ","is_dataset_classified":null,"base_score":1.3862943611198906,"endowment":1.3862943611198906,"datacite_reuse_total":0,"file_count":0,"downloads":0,"views":0,"has_version_chain":false,"is_dataset":false,"is_oa":false,"pmid":"37503140","pmcid":null,"openalex_id":"https://openalex.org/W4384820703","authors":[],"funders":[{"funder_name":"National Institutes of Health","grant_id":"5R01DA041528-05","title":"Using event-related potentials to longitudinally track cue-induced craving incubation in cocaine addicted individuals"}],"total_grants":1,"fwci":null,"citation_percentile":null,"influential_citations":0,"citation_trend":[{"year":2024,"count":1},{"year":2025,"count":2}],"oa_status":"green","license":"cc-by-nc-nd","oa_locations":[{"url":"https://www.biorxiv.org/content/biorxiv/early/2023/07/19/2023.07.18.549548.full.pdf","host_type":"repository"},{"url":"https://www.biorxiv.org/content/biorxiv/early/2023/07/19/2023.07.18.549548.full.pdf","host_type":"repository"},{"url":"https://syndication.highwire.org/content/doi/10.1101/2023.07.18.549548","host_type":"publisher"},{"url":"http://dx.doi.org/10.1101/2023.07.18.549548","host_type":"repository"},{"url":"https://pubmed.ncbi.nlm.nih.gov/37503140","host_type":"repository"},{"url":"https://www.ncbi.nlm.nih.gov/pmc/articles/10370100","host_type":"repository"},{"url":"https://pmc.ncbi.nlm.nih.gov/articles/PMC10370100/pdf/nihpp-2023.07.18.549548v1.pdf","host_type":"repository"},{"url":"https://doi.org/10.1101/2023.07.18.549548","host_type":""}],"fields_of_study":["Mental Health via Writing","Substance Abuse Treatment and Outcomes","Neurotransmitter Receptor Influence on Behavior","03 medical and health sciences","0305 other medical science"],"mesh_terms":[],"keywords":["Abstinence","Term (time)","Cocaine use","Biomarker","Psychology","Drug","Illicit drug","Cognitive psychology","Medicine","Psychiatry","Biology","Article"],"sdg_mappings":[{"sdg_number":0,"sdg_label":"Good health and well-being"}],"linked_datasets":[],"clinical_trials":[],"software_tools":[],"database_accessions":[],"source":"live","citation_network_status":"fetched"},"created_at":"2026-08-04T14:42:38.743543Z","pmid":null,"pmcid":null,"fwci":null,"citation_percentile":null,"influential_citations":0,"oa_status":null,"license":null,"views":0,"total_file_size_bytes":0,"version_count":0,"fair_f":null,"fair_a":null,"fair_i":null,"fair_r":null,"fair_zscore":null,"fair_rationale":null,"fair_model":null,"fair_agent_version":null,"fair_fulltext_source":null,"fair_has_llm":null,"fair_computed_at":null,"clinical_trials":[],"software_tools":[],"db_accessions":[],"linked_datasets":[],"topics":[]}