The purpose of this study is to develop a novel deep-learning-based survival prediction model employing patient activity data recorded by a wearable device.

This study aims to develop a deep-learning-based survival prediction model that utilizes patient movement data upon admission to predict their clinical outcomes: either death or discharge with stable condition. Objective data of the patients are recorded by a wearable device and documented as parameters of physical activity, angle, and spin. In addition to objective data, the investigators also document patients' Karnofsky Performance Status assessed subjectively by clinical doctors. Finally, the investigators aim to explore and describe the applicability, potential, and limitations of the survival prediction model based on patient movement data as a simple prognostic parameter in clinical settings.

Inclusion Criteria:

• Participants aged 20 years or older admitted to the hospice care unit at Taipei Medical University Hospital
• Participants diagnosed with at least one end-stage solid tumor diseases
• Participants consented to receive hospice care

Exclusion Criteria:

• Participants aged below 20 years of age
• Participants diagnosed with leukemia or carcinoma of unknown primary
• Participants with evident signs of approaching death upon admission
• Participants with no vital signs upon admission
• Participants who continued to receive aggressive treatment despite admission to the hospice care unit

Patients aged 20 years or older who were admitted to the hospice care unit at Taipei Medical University Hospital with at least one diagnosis of end-stage solid tumor diseases.