EUVIMEDEUVIMED entdecken
QuerschnittstudieOpen Access · CC BY 4.0

Artificial Intelligence in Classical Homoeopathy: A Cross-Sectional Observational Study on Clinical Utility, Repertorial Accuracy, and Therapeutic Outcome

Dr. Pankaj N. Lathiya, Dr. Zameer F. Refai, Dr. Akshay M. Patel, Dr. Vibhutiben Anilkumar Umaravia, Dr. Pathikkumar Jayeshbhai Patel, Dr. Radha Mangukiya, Dr. Harshna S. Goti

Computer and Decision Making: An International Journal · 03.11.2026 · 3 (6S) · 1268-1280

DOI: 10.1922/comdem.v3i6s.18909 ↗

Kurz gefasst

Querschnittstudie in der Zeitschrift Computer and Decision Making: An International Journal, online erschienen am 03.11.2026. Verfasst von Dr. Pankaj N. Lathiya und 6 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Use of artificial intelligence in classical homoeopathic repertories can help in improving the computational decision making, information retrieval and evidence traceability. But comparative study between lexical, semantic and hierarchy aware retrieval approaches is still not well explored. In this study, an explainable AI-assisted repertorial retrieval and remedy ranking framework was developed and evaluated by combining BM25, sentence embedding using Sentence-BERT, RRF, hierarchical reranking, and evidence-traceable remedy aggregation. There were 74,513 hierarchical rubrics and 507,179 of rubric–remedy occurrences in the Kent Repertory. 300 source-derived queries were used to test seven retrieval configurations. Recall at position K, Mean Reciprocal Rank at position 10 and Normalised Discounted Cumulative Gain at position 10 were used to evaluate the performance of the proposed system, along with paired statistical testing, ablation analysis and robustness evaluation. The BM25–Sentence-BERT hybrid obtained the best NDCG@10 score of 0.9951 and MRR@10 score of 0.9933, which are better than the proposed QHNS-RDS V4 framework (0.9656 and 0.9536, respectively). The result of the statistical comparison showed that the ranking effectiveness of the proposed configuration was significantly low. For NDCG@10, removing hierarchical reranking resulted in a significant improvement of 0.9946, indicating that the fixed structural weighting hurt retrieval performance. All models were negatively affected by severe query truncation, while semantic embeddings made a positive contribution. However, the framework produced remedy ranking results that provided interpretable information using supporting rubric identifiers, hierarchical paths, and source references, and allowed for clear review of computational remedy rankings. These findings prove the feasibility of evidence-traceable repertorial decision support but not necessarily the improvement of clinical decision-making or therapeutic effectiveness. Limitations in generalisability include source-derived benchmarking, OCR uncertainties and inactive uncertainty weighting. The next steps would be to use separately annotated queries, expert verified associations, adaptive-hierarchical scoring and human-centred evaluation for future research prior to practical clinical use.

Zitieren

APA

Lathiya, D. P. N., Refai, D. Z. F., Patel, D. A. M., Umaravia, D. V. A., Patel, D. P. J., Mangukiya, D. R., & Goti, D. H. S. (2026). Artificial Intelligence in Classical Homoeopathy: A Cross-Sectional Observational Study on Clinical Utility, Repertorial Accuracy, and Therapeutic Outcome. Computer and Decision Making: An International Journal, 3(6S), 1268-1280. https://doi.org/10.1922/comdem.v3i6s.18909

Vancouver

Lathiya DPN, Refai DZF, Patel DAM, Umaravia DVA, Patel DPJ, Mangukiya DR, et al. Artificial Intelligence in Classical Homoeopathy: A Cross-Sectional Observational Study on Clinical Utility, Repertorial Accuracy, and Therapeutic Outcome. Computer and Decision Making: An International Journal. 2026;3(6S):1268-1280. doi:10.1922/comdem.v3i6s.18909

BibTeX
@article{Lathiya2026,
  author = {Lathiya, Dr. Pankaj N. and Refai, Dr. Zameer F. and Patel, Dr. Akshay M. and Umaravia, Dr. Vibhutiben Anilkumar and Patel, Dr. Pathikkumar Jayeshbhai and Mangukiya, Dr. Radha and Goti, Dr. Harshna S.},
  title = {Artificial Intelligence in Classical Homoeopathy: A Cross-Sectional Observational Study on Clinical Utility, Repertorial Accuracy, and Therapeutic Outcome},
  journal = {Computer and Decision Making: An International Journal},
  year = {2026},
  volume = {3},
  number = {6S},
  pages = {1268--1280},
  doi = {10.1922/comdem.v3i6s.18909}
}
Kostenlos starten →