Scholarlink

In this paper, machine learning algorithms have used to estimate the length of stay (LOS) of subjects with lower limbs fractures. The dataset used comes from the A.O. Federico II based in Naples and includes 132 patients. KNIME software were used to predict LOS using different type of Machine Learning (ML) classification models. The input variables were sex, age, the presence of comorbidities (0/1), the hospitalization regime (1–5) and mode of discharge (1/2). The very best result was attained with SVM and LR algorithms with an accuracy of 77.78%. These values obtained were compared with those coming from the analysis of data obtained from two other Hospital (“San Giovanni di Dio and Ruggi d'Aragona” of Salerno and at the “A.O.R.N. Antonio Cardarelli” of Naples) to evaluate the difference from the hospitals. For the future development, using the same algorithms, the best efficiency is obtained with the largest dataset.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Classification Algorithms to Predict Hospitalization for a Lower Limb Fracture: A Multicenter Analysis

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Marta Rosaria Marino,
  • Anna Borrelli,
  • Maria Triassi,
  • Giovanni Improta

摘要

In this paper, machine learning algorithms have used to estimate the length of stay (LOS) of subjects with lower limbs fractures. The dataset used comes from the A.O. Federico II based in Naples and includes 132 patients. KNIME software were used to predict LOS using different type of Machine Learning (ML) classification models. The input variables were sex, age, the presence of comorbidities (0/1), the hospitalization regime (1–5) and mode of discharge (1/2). The very best result was attained with SVM and LR algorithms with an accuracy of 77.78%. These values obtained were compared with those coming from the analysis of data obtained from two other Hospital (“San Giovanni di Dio and Ruggi d'Aragona” of Salerno and at the “A.O.R.N. Antonio Cardarelli” of Naples) to evaluate the difference from the hospitals. For the future development, using the same algorithms, the best efficiency is obtained with the largest dataset.