EUVIMEDEUVIMED entdecken
BeobachtungsstudieOpen Access

Assessment of Predictive Factors That Shorten Duration of Treatment in Patients With Multiple Myeloma Using AI: Real-World Longitudinal Study Using Data From Medical Data Vision Claims Database

Hiroshi Handa iD, Tadao Ishida iD, Shuji Ozaki iD, Shinsuke Iida iD, Kittima Wattanakamolkul iD, Chika Sakai iD, Kenichi Kato iD, David Bin-Chia Wu iD, DaeYoung Yu iD, Shota Nemoto iD, Yasuho Yamashita iD, Takuma Shibahara iD

JMIR Cancer · 2026

DOI: 10.2196/75586 ↗

Kurz gefasst

Beobachtungsstudie in der Zeitschrift JMIR Cancer, 2026. Verfasst von Hiroshi Handa und 11 weiteren Autor:innen. Beteiligte Einrichtung: Department of Hematology, Gunma University Hospital, Gunma, Japan. Hauptthemen laut MEDLINE: Multiple Myeloma. Die Arbeit nennt 48 Literaturangaben. Der Volltext ist frei in PubMed Central verfügbar.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Background

With the availability of newer therapies, the duration of therapy (DoT) shortens with each increasing line of treatment in Japanese patients with multiple myeloma (MM).

Objective

This study aimed to identify factors that shorten DoT in patients with MM using a machine learning (ML) procedure from the Medical Data Vision (MDV) database.

Methods

This nationwide, retrospective observational real-world study was conducted using anonymized patient data from the MDV claims database from 2003 to 2022. Patients (≥18 y) with transplant-ineligible newly diagnosed MM (continued first line therapy), or relapsed or refractory MM (continued second or third line therapies) were included. To identify important predictive factors, an explainable deep learning model was created using 647 extracted variables (continuous, binary, and nominal categorical) from the MDV database, and the extracted data were used to train ML algorithms to build point-wise linear (PWL) models for predicting DoT. Predictive performance of the PWL model was compared with the elastic net (regularized logistic regression) and the extreme gradient boosting models, and calculated by area under the curve and evaluated by 10-fold double cross-validation. A clustering analysis (k-means method) of 4848 individual samples assessed the relationship between each sample and DoT (3, 6, and 12 months). The characteristics of clusters and sample features belonging to each cluster during and after treatment were studied.

Results

Overall, 2762 (4848 individual samples) patients were evaluated (mean age 69.6, SD 11.8 years; 1450/2762, 52.5% male). The area under the curve score of the PWL model to predict DoT at 3, 6, and 12 months was 0.61, 0.64, and 0.66, respectively. Based on the similarity of coefficients of regression models, samples were categorized into 2 clusters (clusters A and B) at DoT of 3 months, 3 clusters (clusters A, B, and C) at 6 months, and 12 months (clusters A, B, and C). Cluster B versus cluster A (at 3 months) and cluster C versus cluster A and B (at 6 and 12 months) had a significantly (P

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Multiple MyelomaAgedFemaleHumansMaleMiddle AgedClustering AlgorithmsDatabases, FactualJapanLongitudinal StudiesMachine LearningPredictive Learning ModelsRetrospective StudiesTime Factors

Stichwörter der Autor:innen

Japanartificial intelligencedatabaseduration of therapymachine learningmultiple myelomaprediction probability

Autor:innen und Institutionen

  • Department of Hematology, Gunma University Hospital, Gunma, Japan.
  • Department of Hematology, Japanese Red Cross Medical Center, Tokyo, Japan.
  • Department of Hematology, Tokushima Prefectural Central Hospital, Tokushima, Japan.
  • Department of Hematology and Oncology, Nagoya City University Graduate School of Medical Sciences, Nagoya, Japan.
  • Value, Evidence and Access Department, Integrated Market Access, Johnson & Johnson, Tokyo, Japan.
  • Medical Affairs, Johnson & Johnson, Tokyo, Japan.

ORCID-Profile: Hiroshi Handa, Tadao Ishida, Shuji Ozaki, Shinsuke Iida, Kittima Wattanakamolkul, Chika Sakai, Kenichi Kato, David Bin-Chia Wu, DaeYoung Yu, Shota Nemoto, Yasuho Yamashita, Takuma Shibahara

Zitieren

APA

Handa, H., Ishida, T., Ozaki, S., Iida, S., Wattanakamolkul, K., Sakai, C., Kato, K., Bin-Chia Wu, D., Yu, D., Nemoto, S., Yamashita, Y., & Shibahara, T. (2026). Assessment of Predictive Factors That Shorten Duration of Treatment in Patients With Multiple Myeloma Using AI: Real-World Longitudinal Study Using Data From Medical Data Vision Claims Database. JMIR Cancer. https://doi.org/10.2196/75586

Vancouver

Handa H, Ishida T, Ozaki S, Iida S, Wattanakamolkul K, Sakai C, et al. Assessment of Predictive Factors That Shorten Duration of Treatment in Patients With Multiple Myeloma Using AI: Real-World Longitudinal Study Using Data From Medical Data Vision Claims Database. JMIR Cancer. 2026. doi:10.2196/75586

BibTeX
@article{Handa2026,
  author = {Handa, Hiroshi and Ishida, Tadao and Ozaki, Shuji and Iida, Shinsuke and Wattanakamolkul, Kittima and Sakai, Chika and Kato, Kenichi and Bin-Chia Wu, David and Yu, DaeYoung and Nemoto, Shota and Yamashita, Yasuho and Shibahara, Takuma},
  title = {Assessment of Predictive Factors That Shorten Duration of Treatment in Patients With Multiple Myeloma Using AI: Real-World Longitudinal Study Using Data From Medical Data Vision Claims Database},
  journal = {JMIR Cancer},
  year = {2026},
  doi = {10.2196/75586}
}
Kostenlos starten →