NLP Analysis of Weekly Narratives for Dynamic Clinical Assessment in SUD (DYNA-NLP)

Analysis of Weekly Narratives Using Natural Language Processing in Patients Undergoing Residential Rehabilitation: A Hybrid Approach for the Dynamic Assessment of Clinical Processes

This prospective observational study follows adults undergoing residential rehabilitation for severe substance use disorders at a specialized treatment center in Mexico. Participants provide weekly written narratives describing their emotions, challenges, coping strategies, and treatment experiences, and complete validated psychological questionnaires every two weeks, including the Generalized Anxiety Disorder-7 (GAD-7), Environmental Reward Observation Scale (EROS), Automatic Thoughts Questionnaire-8 (ATQ-8), and Behavioral Activation for Depression Scale (BADS).

The study applies natural language processing (NLP) and machine learning methods to analyze participants' narratives and identify emotional, cognitive, and behavioral patterns associated with clinical change over time. Narrative-derived features are combined with questionnaire scores to generate a dynamic clinical risk representation that may help detect early signs of psychological worsening or improvement during residential treatment.

Participants continue receiving standard residential care, and the study does not modify treatment decisions or clinical interventions. Up to 35 participants with sufficient longitudinal follow-up data will be included in the primary analysis. Data collection is expected to continue through September 2026.

Study Overview

• Status: Recruiting
• Conditions: Anxiety; Addiction; Emotional Dysregulation; Rumination - Thoughts; Substance Use Disorder (SUD)
• Intervention / Treatment: Other: Residential rehabilitation as usual

Detailed Description

Patients undergoing treatment for substance use disorders frequently describe changes in mood, motivation, hopelessness, cognitive rigidity, emotional distress, and coping strategies through spontaneous written language. These narratives may contain clinically meaningful indicators associated with psychological deterioration, treatment progress, or relapse vulnerability. However, systematic manual analysis of longitudinal written narratives is difficult to implement in routine clinical practice because of the volume and complexity of the data.

Recent developments in natural language processing (NLP), representation learning, and deep learning provide methods for extracting quantitative linguistic and semantic information from written text. Integrating these features with repeated psychometric assessments may support the development of longitudinal models capable of characterizing changes in mental health status over time in patients receiving residential addiction treatment.

Objectives

The objectives of this study are:

To extract semantic, emotional, and linguistic features from weekly patient narratives using NLP methods, including sentence embeddings, sentiment and emotion classification, and semantic similarity analyses based on Acceptance and Commitment Therapy (ACT) constructs.

To integrate narrative-derived variables with repeated psychometric measures (Generalized Anxiety Disorder-7 (GAD-7), Environmental Reward Observation Scale (EROS), Automatic Thoughts Questionnaire-8 (ATQ-8), and Behavioral Activation for Depression Scale (BADS)) using a multimodal deep learning autoencoder capable of generating a low-dimensional representation of longitudinal clinical status.

To evaluate whether latent representations generated by the autoencoder are associated with periods of clinical worsening or clinical improvement across time using leave-one-patient-out cross-validation procedures.

To develop a dynamic longitudinal risk representation capable of estimating future changes in automatic negative thoughts, measured through subsequent ATQ-8 scores.

Study Design

This study is a prospective observational cohort conducted at a single residential rehabilitation center in Mexico.

Eligible participants are adults aged 18 years or older with a diagnosis of severe substance use disorder who are admitted for residential treatment. Individuals with active psychotic symptoms or cognitive impairment that substantially interferes with the ability to complete written narratives are excluded.

Participants complete:

Weekly digital written narratives using open-ended prompts focused on emotions, challenges, coping responses, interpersonal experiences, and perceived treatment progress.

Biweekly administration of the following validated self-report instruments:

GAD-7

EROS

ATQ-8

BADS

All information is collected through digital forms integrated into routine clinical monitoring procedures at the treatment center.

Participants may contribute data for up to 20 weeks, depending on duration of residential stay. The study began on 25 May 2025, and primary completion is anticipated in September 2026.

Up to 35 participants with sufficient longitudinal observations will be included in the primary analytic cohort.

NLP and Machine Learning Pipeline

Text preprocessing

Narratives are written in Spanish and undergo preprocessing procedures that include:

minimum length filtering, normalization, consolidation of narrative fields into a single weekly text sample, tokenization and linguistic annotation.

Feature extraction

Narrative features include:

Sentiment classification (positive, neutral, negative) Emotion probabilities for joy, sadness, anger, fear, surprise, and disgust using the pysentimiento library

Linguistic variables including:

type-token ratio, mean sentence length, proportion of first-person pronouns, proportion of past-tense verbs, obtained through udpipe

Semantic similarity measures between patient narratives and ACT-related prototype domains including:

experiential avoidance, cognitive fusion, rule-governed behavior, helplessness, achievement orientation, hopefulness

Semantic similarity is computed through cosine similarity between sentence-transformer embeddings and predefined prototype centroids.

Dimensionality reduction procedures using principal component analysis (PCA) are applied to selected linguistic and semantic variables to derive a principal component representing orientation toward internal emotional experience versus external contextual events.

Autoencoder Architecture

The multimodal model receives concatenated longitudinal feature vectors after robust scaling.

The architecture includes:

a bidirectional long short-term memory (BiLSTM) encoder, multi-head attention mechanisms, variational regularization using a β-variational autoencoder (β-VAE) framework, a decoder trained to reconstruct the original temporal feature sequence, and an auxiliary classification component predicting next-period clinical worsening.

The training objective combines:

mean squared reconstruction error, focal loss for classification, Kullback-Leibler divergence, and optional temporal smoothness regularization.

Model performance is evaluated using leave-one-patient-out cross-validation procedures. A final model may subsequently be trained using the complete dataset.

Primary Analyses

Primary analyses include:

evaluation of discriminative performance for prediction of subsequent clinical worsening using: area under the receiver operating characteristic curve (AUROC), area under the precision-recall curve (AUPRC), F1 score, Matthews correlation coefficient (MCC) examination of associations between latent trajectory representations and longitudinal psychometric changes, development of a Composite Clinical Risk Index (CCRI) derived from latent representations, and comparison of CCRI trajectories with weeks classified as clinical worsening according to predefined multimodal criteria.

Ethics and Dissemination

The study has received approval from the institutional review board of Under The Tree Miller A.C.

All participants provide written informed consent prior to participation.

Participation does not alter or replace standard residential treatment. All clinical decisions remain under the responsibility of treating professionals independent of study procedures.

Study findings will be submitted for publication in peer-reviewed scientific journals regardless of outcome. De-identified datasets and analysis code may be made available upon reasonable request and in accordance with institutional and ethical requirements.

Recruitment Status

Recruitment is ongoing. Final data collection for the primary outcome is anticipated in September 2026.

• Study Type: Observational
• Enrollment (Estimated): 35

Contacts and Locations

• Study Contact: Name: Ricardo Fernandez; Phone Number: +52 1 33 1544 5474
• Study Contact Backup: Name: Lauro Gutiérrez Castro; Phone Number: +523314696107; Email: saraqael_sefer@hotmail.com

Study Locations

• Mexico
  • Jalisco
    • Potrerillos, Jalisco, Mexico, 45815
      • Recruiting
      • Under The Tree Potrerillos
      • Contact: Lauro Gutiérrez Castro; Phone Number: +523314696107; Email: saraqael_sefer@hotmail.com

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: Adult; Older Adult
• Accepts Healthy Volunteers: No

• Sampling Method: Non-Probability Sample

Study Population

Adult males (≥18 years) with severe substance use disorder, residing in a single residential rehabilitation center in Mexico. Participants are enrolled consecutively as they enter the program and meet eligibility criteria. The sample is non-probabilistic, reflecting the real-world clinical population of this specific center.

Description

Inclusion Criteria:

1. Clinical diagnosis of severe substance use disorder (polydrug use, including cocaine, methamphetamines, alcohol, and/or cannabis), confirmed by the center's admission assessment.
2. Male sex (all participants in the center's residential program are male).
3. Age 18 years or older.
4. Current resident of the participating residential rehabilitation center in Mexico.
5. Completed at least four weeks of residential treatment at the time of study enrollment.
6. Able to write coherent weekly narratives in Spanish (no severe cognitive impairment or active psychosis).
7. Willing to provide written informed consent.

Exclusion Criteria:

1. Presence of acute psychotic symptoms that interfere with the ability to write or understand the study procedures.
2. Severe cognitive impairment (e.g., due to traumatic brain injury, intellectual disability) that prevents meaningful narrative production.
3. Inability to comply with weekly narrative writing (e.g., illiteracy, severe visual impairment).
4. Planned discharge from the residential program within less than 4 weeks from enrollment.
5. Enrollment in another interventional clinical trial that could confound the interpretation of outcomes.

Study Plan

How is the study designed?

Number of groups / cohorts

1

Cohorts and Interventions

Group / Cohort

Intervention / Treatment

Residential rehabilitation cohort; Adult men (≥18 years) with severe substance use disorder, admitted to a residential rehabilitation center in Mexico. All participants receive the center's standard, multimodal treatment program, which includes group therapy, individual counseling, occupational activities, 12-step facilitation, and relapse prevention education. The study does not assign, modify, or withhold any component of this program. Participants are followed prospectively for the duration of their stay (10-20 weeks).

Other: Residential rehabilitation as usual; Participants follow the standard residential rehabilitation program provided by the center. This is a naturalistic exposure; the study does not impose any additional intervention. The "intervention" of interest is the routine therapeutic environment and its associated psychological processes (e.g., changes in avoidance, cognitive fusion, hope, and behavioral activation). These processes are measured through weekly written narratives and biweekly validated clinical scales (GAD-7, EROS, ATQ-8, BADS).

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Change in negative automatic thoughts measured by the Automatic Thoughts Questionnaire 8-item version	The Automatic Thoughts Questionnaire 8-item version is administered every two weeks. Total scores range from 8 to 40, with higher scores indicating more frequent negative automatic thoughts. The primary outcome is the change from baseline in total score over the course of residential treatment.	Every 2 weeks from baseline until discharge from residential treatment (up to 20 weeks)

Secondary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Change in generalized anxiety symptoms measured by the Generalized Anxiety Disorder 7 item scale	The Generalized Anxiety Disorder 7-item scale is administered every two weeks. Total scores range from 0 to 21, with higher scores indicating greater anxiety symptom severity.	Every 2 weeks from baseline until discharge (up to 20 weeks)
Change in environmental reward measured by the Environmental Reward Observation Scale	The Environmental Reward Observation Scale iis administered every two weeks. Total scores range from 10 to 40, with higher scores indicating greater exposure to positive environmental reinforcement.	Every 2 weeks from baseline until discharge (up to 20 weeks)
Change in behavioral activation and avoidance measured by the Behavioral Activation for Depression Scale	The Behavioral Activation for Depression Scale is administered every two weeks. Higher scores on the Activation subscale indicate greater engagement in goal-directed activity; higher scores on the Avoidance/Rumination subscale indicate greater behavioral avoidance and rumination.	Every 2 weeks from baseline until discharge (up to 20 weeks)
Weekly self-reported emotional intensity	Participants report the intensity of their predominant weekly emotion on a 0 to 10 numeric rating scale, where higher scores indicate greater emotional intensity.	Weekly from baseline until discharge (up to 20 weeks)
Weekly self-reported craving intensity	Participants rate average craving for substance use during the prior week on a 0 to 10 numeric rating scale, where higher scores indicate greater craving intensity.	Weekly from baseline until discharge (up to 20 weeks)

Other Outcome Measures

Outcome Measure	Measure Description	Time Frame
Narrative sentiment valence index derived from weekly written narratives	Sentiment valence is calculated from weekly written narratives using a validated Spanish-language sentiment classification model. Scores range from -1 to +1, with lower scores indicating more negative emotional valence.	Weekly from baseline until discharge (up to 20 weeks)
Semantic similarity to Acceptance and Commitment Therapy-related constructs	Semantic similarity scores are computed between weekly narratives and prototype domains (e.g., experiential avoidance, cognitive fusion) using sentence embedding methods. Scores range from 0 to 1 (higher = greater similarity).	Weekly from baseline until discharge (up to 20 weeks)
Composite Clinical Risk Index derived from longitudinal narrative features	A composite longitudinal risk score is a composite longitudinal risk score calculated using multiple narrative-derived emotional variability indicators. Higher scores indicate greater estimated clinical vulnerability over time.	Weekly from baseline until discharge (up to 20 weeks)
Linguistic markers extracted from weekly written narratives	Linguistic variables including lexical diversity (type-token ratio), first-person pronoun frequency, and past-tense verb proportion are extracted from weekly written narratives using automated natural language processing (NLP) procedures.	Weekly from baseline until discharge (up to 20 weeks)

Collaborators and Investigators

• Sponsor: Lauro Gutiérrez Castro
• Investigators: Principal Investigator: Lauro Gutiérrez Castro, Under The Tree

Study record dates

Study Major Dates

• Study Start (Actual): May 25, 2025
• Primary Completion (Estimated): September 1, 2026
• Study Completion (Estimated): September 1, 2026

Study Registration Dates

• First Submitted: May 9, 2026
• First Submitted That Met QC Criteria: May 14, 2026
• First Posted (Actual): May 19, 2026

Study Record Updates

• Last Update Posted (Actual): May 19, 2026
• Last Update Submitted That Met QC Criteria: May 14, 2026
• Last Verified: May 1, 2026

More Information

Terms related to this study

• Keywords: Behavioral Activation; Natural Language Processing; Machine Learning; Automatic Thoughts; Residential Rehabilitation
• Additional Relevant MeSH Terms: Mental Disorders; Chemically-Induced Disorders; Compulsive Behavior; Impulsive Behavior; Behavior; Anxiety Disorders; Substance-Related Disorders; Behavior, Addictive
• Other Study ID Numbers: SISAP-TUS-HYB/WN-01

Plan for Individual participant data (IPD)

• Plan to Share Individual Participant Data (IPD)?: YES
• IPD Plan Description: De-identified individual participant data (IPD) from the weekly narratives and biweekly clinical scales, after removal of all direct identifiers (name, date of birth, exact dates converted to study day/week numbers). Only aggregated or pseudonymized data will be shared.
• IPD Sharing Time Frame: IPD and supporting information will be available starting 6 months after publication of the primary results and will remain available for 5 years.
• IPD Sharing Access Criteria: Data will be shared upon reasonable request to the corresponding author. Requesters must sign a data use agreement that prohibits re-identification and restricts use to replicating the published analyses or conducting secondary analyses approved by the study's ethics committee. De-identified data will be provided in CSV format; analytic code will be provided as R and Python scripts via a public repository (e.g., GitHub).
• IPD Sharing Supporting Information Type: STUDY_PROTOCOL; SAP; ICF; ANALYTIC_CODE

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No