Automated evaluation of psychotherapy skills using speech and language technologies.

With the growing prevalence of psychological interventions, it is vital to have measures which rate the effectiveness of psychological care to assist in training, supervision, and quality assurance of services. Traditionally, quality assessment is addressed by human raters who evaluate recorded sessions along specific dimensions, often codified through constructs relevant to the approach and domain. This is, however, a cost-prohibitive and time-consuming method that leads to poor feasibility and limited use in real-world settings. To facilitate this process, we have developed an automated competency rating tool able to process the raw recorded audio of a session, analyzing who spoke when, what they said, and how the health professional used language to provide therapy. Focusing on a use case of a specific type of psychotherapy called "motivational interviewing", our system gives comprehensive feedback to the therapist, including information about the dynamics of the session (e.g., therapist's vs. client's talking time), low-level psychological language descriptors (e.g., type of questions asked), as well as other high-level behavioral constructs (e.g., the extent to which the therapist understands the clients' perspective). We describe our platform and its performance using a dataset of more than 5000 recordings drawn from its deployment in a real-world clinical setting used to assist training of new therapists. Widespread use of automated psychotherapy rating tools may augment experts' capabilities by providing an avenue for more effective training and skill improvement, eventually leading to more positive clinical outcomes.

Machine learning and natural language processing in psychotherapy research: Alliance as example use case.

Artificial intelligence generally and machine learning specifically have become deeply woven into the lives and technologies of modern life. Machine learning is dramatically changing scientific research and industry and may also hold promise for addressing limitations encountered in mental health care and psychotherapy. The current paper introduces machine learning and natural language processing as related methodologies that may prove valuable for automating the assessment of meaningful aspects of treatment. Prediction of therapeutic alliance from session recordings is used as a case in point. Recordings from 1,235 sessions of 386 clients seen by 40 therapists at a university counseling center were processed using automatic speech recognition software. Machine learning algorithms learned associations between client ratings of therapeutic alliance exclusively from session linguistic content. Using a portion of the data to train the model, machine learning algorithms modestly predicted alliance ratings from session content in an independent test set (Spearman's ρ = .15, p < .001). These results highlight the potential to harness natural language processing and machine learning to predict a key psychotherapy process variable that is relatively distal from linguistic content. Six practical suggestions for conducting psychotherapy research using machine learning are presented along with several directions for future research. Questions of dissemination and implementation may be particularly important to explore as machine learning improves in its ability to automate assessment of psychotherapy process and outcome. (PsycInfo Database Record (c) 2020 APA, all rights reserved).

The Silent Treatment?: Changes in patient emotional expression after silence.

Psychotherapy can be an emotionally laden conversation, where both verbal and non-verbal interventions may impact the therapeutic process. Prior research has postulated mixed results in how clients emotionally react following a silence after the therapist is finished talking, potentially due to studying a limited range of silences with primarily qualitative and self-report methodologies. A quantitative exploration may illuminate new findings. Utilizing research and automatic data processing from the field of linguistics, we analysed the full range of silence lengths (0.2 to 24.01 seconds), and measures of emotional expression - vocally encoded arousal and emotional valence from the works spoken - of 84 audio recordings Motivational Interviewing sessions. We hypothesized that both the level and the variance of client emotional expression would change as a function of silence length, however, due to the mixed results in the literature the direction of emotional change was unclear. We conducted a multilevel linear regression to examine how the level of client emotional expression changed across silence length, and an ANOVA to examine the variability of client emotional expression across silence lengths. Results indicated in both analyses that as silence length increased, emotional expression largely remained the same. Broadly, we demonstrated a weak connection between silence length and emotional expression, indicating no persuasive evidence that silence leads to client emotional processing and expression.

A Machine-Learning Algorithm for the Automated Perceptual Evaluation of Dysphonia Severity.

OBJECTIVES: Auditory-perceptual assessments are the gold standard for assessing voice quality. This project aims to develop a machine-learning model for measuring perceptual dysphonia severity of audio samples consistent with assessments by expert raters. METHODS: The Perceptual Voice Qualities Database samples were used, including sustained vowel and Consensus Auditory-Perceptual Evaluation of Voice sentences, which were previously expertly rated on a 0-100 scale. The OpenSMILE (audEERING GmbH, Gilching, Germany) toolkit was used to extract acoustic (Mel-Frequency Cepstral Coefficient-based, n = 1428) and prosodic (n = 152) features, pitch onsets, and recording duration. We utilized a support vector machine and these features (n = 1582) for automated assessment of dysphonia severity. Recordings were separated into vowels (V) and sentences (S) and features were extracted separately from each. Final voice quality predictions were made by combining the features extracted from the individual components with the whole audio (WA) sample (three file sets: S, V, WA). RESULTS: This algorithm has a high correlation (r = 0.847) with estimates of expert raters. The root mean square error was 13.36. Increasing signal complexity resulted in better estimation of dysphonia, whereby combining the features outperformed WA, S, and V sets individually. CONCLUSION: A novel machine-learning algorithm was able to perform perceptual estimates of dysphonia severity using standardized audio samples on a 100-point scale. This was highly correlated to expert raters. This suggests that ML algorithms could offer an objective method for evaluating voice samples for dysphonia severity.

A machine learning approach to improve implementation monitoring of family-based preventive interventions in primary care.

Background: Evidence-based parenting programs effectively prevent the onset and escalation of child and adolescent behavioral health problems. When programs have been taken to scale, declines in the quality of implementation diminish intervention effects. Gold-standard methods of implementation monitoring are cost-prohibitive and impractical in resource-scarce delivery systems. Technological developments using computational linguistics and machine learning offer an opportunity to assess fidelity in a low burden, timely, and comprehensive manner. Methods: In this study, we test two natural language processing (NLP) methods [i.e., Term Frequency-Inverse Document Frequency (TF-IDF) and Bidirectional Encoder Representations from Transformers (BERT)] to assess the delivery of the Family Check-Up 4 Health (FCU4Health) program in a type 2 hybrid effectiveness-implementation trial conducted in primary care settings that serve primarily Latino families. We trained and evaluated models using 116 English and 81 Spanish-language transcripts from the 113 families who initiated FCU4Health services. We evaluated the concurrent validity of the TF-IDF and BERT models using observer ratings of program sessions using the COACH measure of competent adherence. Following the Implementation Cascade model, we assessed predictive validity using multiple indicators of parent engagement, which have been demonstrated to predict improvements in parenting and child outcomes. Results: Both TF-IDF and BERT ratings were significantly associated with observer ratings and engagement outcomes. Using mean squared error, results demonstrated improvement over baseline for observer ratings from a range of 0.83-1.02 to 0.62-0.76, resulting in an average improvement of 24%. Similarly, results demonstrated improvement over baseline for parent engagement indicators from a range of 0.81-27.3 to 0.62-19.50, resulting in an approximate average improvement of 18%. Conclusions: These results demonstrate the potential for NLP methods to assess implementation in evidence-based parenting programs delivered at scale. Future directions are presented. Trial registration: NCT03013309 ClinicalTrials.gov.

Research has shown that evidence-based parenting programs effectively prevent the onset and escalation of child and adolescent behavioral health problems. However, if they are not implemented with fidelity, there is a potential that they will not produce the same effects. Gold-standard methods of implementation monitoring include observations of program sessions. This is expensive and difficult to implement in delivery settings with limited resources. Using data from a trial of the Family Check-Up 4 Health program in primary care settings that served Latino families, we investigated the potential to make use of a form of machine learning called natural language processing (NLP) to monitor program delivery. NLP-based ratings were significantly associated with independent observer ratings of fidelity and participant engagement outcomes. These results demonstrate the potential for NLP methods to monitor implementation in evidence-based parenting programs delivered at scale.

An Automated Quality Evaluation Framework of Psychotherapy Conversations with Local Quality Estimates.

Text-based computational approaches for assessing the quality of psychotherapy are being developed to support quality assurance and clinical training. However, due to the long durations of typical conversation based therapy sessions, and due to limited annotated modeling resources, computational methods largely rely on frequency-based lexical features or dialogue acts to assess the overall session level characteristics. In this work, we propose a hierarchical framework to automatically evaluate the quality of transcribed Cognitive Behavioral Therapy (CBT) interactions. Given the richly dynamic nature of the spoken dialog within a talk therapy session, to evaluate the overall session level quality, we propose to consider modeling it as a function of local variations across the interaction. To implement that empirically, we divide each psychotherapy session into conversation segments and initialize the segment-level qualities with the session-level scores. First, we produce segment embeddings by fine-tuning a BERT-based model, and predict segment-level (local) quality scores. These embeddings are used as the lower-level input to a Bidirectional LSTM-based neural network to predict the session-level (global) quality estimates. In particular, we model the global quality as a linear function of the local quality scores, which allows us to update the segment-level quality estimates based on the session-level quality prediction. These newly estimated segment-level scores benefit the BERT fine-tuning process, which in turn results in better segment embeddings. We evaluate the proposed framework on automatically derived transcriptions from real-world CBT clinical recordings to predict session-level behavior codes. The results indicate that our approach leads to improved evaluation accuracy for most codes when used for both regression and classification tasks.

Automated quality assessment of cognitive behavioral therapy sessions through highly contextualized language representations.

During a psychotherapy session, the counselor typically adopts techniques which are codified along specific dimensions (e.g., 'displays warmth and confidence', or 'attempts to set up collaboration') to facilitate the evaluation of the session. Those constructs, traditionally scored by trained human raters, reflect the complex nature of psychotherapy and highly depend on the context of the interaction. Recent advances in deep contextualized language models offer an avenue for accurate in-domain linguistic representations which can lead to robust recognition and scoring of such psychotherapy-relevant behavioral constructs, and support quality assurance and supervision. In this work, we propose a BERT-based model for automatic behavioral scoring of a specific type of psychotherapy, called Cognitive Behavioral Therapy (CBT), where prior work is limited to frequency-based language features and/or short text excerpts which do not capture the unique elements involved in a spontaneous long conversational interaction. The model focuses on the classification of therapy sessions with respect to the overall score achieved on the widely-used Cognitive Therapy Rating Scale (CTRS), but is trained in a multi-task manner in order to achieve higher interpretability. BERT-based representations are further augmented with available therapy metadata, providing relevant non-linguistic context and leading to consistent performance improvements. We train and evaluate our models on a set of 1,118 real-world therapy sessions, recorded and automatically transcribed. Our best model achieves an F1 score equal to 72.61% on the binary classification task of low vs. high total CTRS.

Feature Fusion Strategies for End-to-End Evaluation of Cognitive Behavior Therapy Sessions.

Cognitive Behavioral Therapy (CBT) is a goal-oriented psychotherapy for mental health concerns implemented in a conversational setting. The quality of a CBT session is typically assessed by trained human raters who manually assign pre-defined session-level behavioral codes. In this paper, we develop an end-to-end pipeline that converts speech audio to diarized and transcribed text and extracts linguistic features to code the CBT sessions automatically. We investigate both word-level and utterance-level features and propose feature fusion strategies to combine them. The utterance level features include dialog act tags as well as behavioral codes drawn from another well-known talk psychotherapy called Motivational Interviewing (MI). We propose a novel method to augment the word-based features with the utterance level tags for subsequent CBT code estimation. Experiments show that our new fusion strategy outperforms all the studied features, both when used individually and when fused by direct concatenation. We also find that incorporating a sentence segmentation module can further improve the overall system given the preponderance of multi-utterance conversational turns in CBT sessions.

ROLE SPECIFIC LATTICE RESCORING FOR SPEAKER ROLE RECOGNITION FROM SPEECH RECOGNITION OUTPUTS.

The language patterns followed by different speakers who play specific roles in conversational interactions provide valuable cues for the task of Speaker Role Recognition (SRR). Given the speech signal, existing algorithms typically try to find such patterns in the output of an Automatic Speech Recognition (ASR) system. In this work we propose an alternative way of revealing role-specific linguistic characteristics, by making use of role-specific ASR outputs, which are built by suitably rescoring the lattice produced after a first pass of ASR decoding. That way, we avoid pruning the lattice too early, eliminating the potential risk of information loss.

Identifying Therapist and Client Personae for Therapeutic Alliance Estimation.

Psychotherapy, from a narrative perspective, is the process in which a client relates an on-going life-story to a therapist. In each session, a client will recount events from their life, some of which stand out as more significant than others. These significant stories can ultimately shape one's identity. In this work we study these narratives in the context of therapeutic alliance-a self-reported measure on the perception of a shared bond between client and therapist. We propose that alliance can be predicted from the interactions between certain types of clients with types of therapists. To validate this method, we obtained 1235 transcribed sessions with client-reported alliance to train an unsupervised approach to discover groups of therapists and clients based on common types of narrative characters, or personae. We measure the strength of the relation between personae and alliance in two experiments. Our results show that (1) alliance can be explained by the interactions between the discovered character types, and (2) models trained on therapist and client personae achieve significant performance gains compared to competitive supervised baselines. Finally, exploratory analysis reveals important character traits that lead to an improved perception of alliance.

Using Prosodic and Lexical Information for Learning Utterance-level Behaviors in Psychotherapy.

In this paper, we present an approach for predicting utterance level behaviors in psychotherapy sessions using both speech and lexical features. We train long short term memory (LSTM) networks with an attention mechanism using words, both manually and automatically transcribed, and prosodic features, at the word level, to predict the annotated behaviors. We demonstrate that prosodic features provide discriminative information relevant to the behavior task and show that they improve prediction when fused with automatically derived lexical features. Additionally, we investigate the weights of the attention mechanism to determine words and prosodic patterns which are of importance to the behavior prediction task.