Natural language processing-based algorithmic decision support system in identifying psychological conflicts using...

Thematic apperception test (TAT) is a personality assessment measure widely applied in various clinical conditions. By analyzing written stories, clinicians can unveil the unconscious dynamics of the mind, including unconscious needs, presses, emotions, and ego defense mechanisms. Natural language processing (NLP) encompasses text feature extraction techniques, such as tokenization, stemming, lemmatization, and removal of stopwords. Recently, NLP-based methods have gained popularity among researchers for analyzing psychological attributes and predicting human behavior. As the TAT is one of the most extensively utilized projective psychological tests with many clinical applications, it is pertinent to explore the potential impact of NLP on its scoring process.

To assess if NLP-based algorithmic decision support can be used to accurately and precisely score across psychometric dimensions of TAT.

An NLP-based algorithmic decision support system (NBADSS) was initially designed using the Python-based library "nltk." Subsequently, 87 healthy participants aged 17-40 years were selected using purposive sampling. Personal history and background information were collected from each participant before they underwent psychometric assessments using the TAT. The TAT protocols were manually analyzed and then analyzed with the help of NBADSS by another researcher. Lastly, the scoring procedures' findings were compared and statistically analyzed using Cohen's Kappa test and ROC curve analysis.

The findings suggest that NBADSS is a valuable adjunct to quantifying TAT protocols objectively. NBADSS is found to be highly sensitive and specific in assessing needs, pressures, emotions, and defense mechanisms present in the stories.

NBADSS is an objective means of uncovering underlying intrapsychic conflicts projected through TAT stories.