TY - JOUR

T1 - Применение методов анализа естественного языка для оценки качества медицинских услуг на основе онлайн-отзывов пациентов

AU - Dudina, VI

AU - Karelin, AA

PY - 2026/6/30

Y1 - 2026/6/30

N2 - Today survey methods for assessing patient satisfaction with medical services are supplemented by the analysis of unstructured online reviews, which expands the empirical research base. Such data sources have complementary functions: surveys provide standardized data, while text reviews reflect the context of patients' interaction with the healthcare organization, emotional and organizational aspects of care. Thus, we need methodologically sound approaches that integrate sociological theories of patient satisfaction with natural language analysis in the assessment of medical services and provide data comparable to survey methods. The study aims at developing and testing an integrative methodology for assessing the quality of healthcare services through the analysis of online patient reviews based on an appropriate theoretical model. The authors conducted an analysis of 26,000 online public reviews of 120 healthcare organizations in Saint Petersburg: texts underwent preprocessing; contextual embeddings from ruRoBERTa were used together with the BERTopic algorithm for topic modeling. For external validation, integrated rankings of healthcare organizations were compared with indicators derived from the survey data. The proposed theoretical model integrated the following key approaches to the healthcare quality assessment: "structure-process-outcome" concept; expectation-disconfirmation theory; "zone of tolerance" model; service quality model; SERVQUAL and HEALTHQUAL models. The analysis revealed moderate but statistically significant consistency between aggregated indicators based on textual reviews and healthcare quality assessments based on survey data, and thematic analysis allowed to identify the most frequent sources of patient dissatisfaction (organizational, communicational, and infrastructural). Positive reviews most frequently focus on the medical staff professionalism, latest clinic equipment, comfortable waiting conditions, and transparency of organizational processes.

AB - Today survey methods for assessing patient satisfaction with medical services are supplemented by the analysis of unstructured online reviews, which expands the empirical research base. Such data sources have complementary functions: surveys provide standardized data, while text reviews reflect the context of patients' interaction with the healthcare organization, emotional and organizational aspects of care. Thus, we need methodologically sound approaches that integrate sociological theories of patient satisfaction with natural language analysis in the assessment of medical services and provide data comparable to survey methods. The study aims at developing and testing an integrative methodology for assessing the quality of healthcare services through the analysis of online patient reviews based on an appropriate theoretical model. The authors conducted an analysis of 26,000 online public reviews of 120 healthcare organizations in Saint Petersburg: texts underwent preprocessing; contextual embeddings from ruRoBERTa were used together with the BERTopic algorithm for topic modeling. For external validation, integrated rankings of healthcare organizations were compared with indicators derived from the survey data. The proposed theoretical model integrated the following key approaches to the healthcare quality assessment: "structure-process-outcome" concept; expectation-disconfirmation theory; "zone of tolerance" model; service quality model; SERVQUAL and HEALTHQUAL models. The analysis revealed moderate but statistically significant consistency between aggregated indicators based on textual reviews and healthcare quality assessments based on survey data, and thematic analysis allowed to identify the most frequent sources of patient dissatisfaction (organizational, communicational, and infrastructural). Positive reviews most frequently focus on the medical staff professionalism, latest clinic equipment, comfortable waiting conditions, and transparency of organizational processes.

KW - healthcare service quality

KW - natural language processing methods

KW - online reviews

KW - patient satisfaction

KW - sentiment analysis

KW - sociology of health

KW - survey

KW - topic modeling

KW - unstructured textual data

UR - https://elibrary.ru/ZKBGZT

UR - https://www.mendeley.com/catalogue/f42a9da3-5d34-3889-8964-3b29585e1e9d/

U2 - 10.22363/2313-2272-2026-26-2-394-412

DO - 10.22363/2313-2272-2026-26-2-394-412

M3 - статья

VL - 26

SP - 394

EP - 412

JO - ВЕСТНИК РОССИЙСКОГО УНИВЕРСИТЕТА ДРУЖБЫ НАРОДОВ. СЕРИЯ: СОЦИОЛОГИЯ

JF - ВЕСТНИК РОССИЙСКОГО УНИВЕРСИТЕТА ДРУЖБЫ НАРОДОВ. СЕРИЯ: СОЦИОЛОГИЯ

SN - 2313-2272

IS - 2

ER -