To study the potential of a predictive Artificial Intelligence tool to foster early identification of symptoms and criticality in breast and cervical cancers in Indian patients.

Artificial intelligence models can help distinguish urgent problems, an especially important goal in overstretched health systems. Timely, effective care of symptoms reduces mortality and improves the quality of life in cancer patients. A predictive model can help predict symptoms and symptomatic criticality earlier in the course of disease. Our overarching objective is to study the potential of a predictive A.I. tool to foster early identification of symptoms and criticality in cervical and breast cancers in Indian female patients. To develop and pilot an AI tool and its outcome to identify critical burdens of patients with breast and cervical cancers the patients with breast and cervical cancers coming to the pain and palliative care OPD/IPD will be enrolled and subsequent follow-up will be done. The enrolment of patients will be done through Google form which consists of Socio-demographic data, WHO well-being questionnaire, ECOG, and ESAS. To ensure best practices for transparency, validation, and reproducibility, we will follow the clinical A.I. modeling checklist and CONSORT-AI standards. Models will be developed through the following reproducible pipeline of 5 steps- 1-Partitioning data, 2-Optimization, 3-Model selection, 4-Performance evaluation, and 5-Model examination. To evaluate clinician experience with the tool and barriers and facilitators to the implementation of the tool and palliative care intervention, the study will be a qualitative study using implementation science methods, which we will pursue by interviewing site clinicians regarding how they might use a mHealth tool and integrate it within their routine workflows. We will identify initial clinical participants at the initial 10 intervention sites by referral from each site’s Principal Investigator. Participants may be multidisciplinary palliative care or oncology team members including physicians and non-physicians with symptom management responsibilities, primarily physicians or nurses. They need to be English (interview by Stanford or AIIMS) or Hindi speakers (interview by AIIMS). Data analysis will follow standard mixed inductive, deductive qualitative methods (e.g., dual coding, development of a primary codebook, secondary coding, iterative thematic development by consensus, goal standard adjudication of disagreement). Atlas.ti Software to be used for data analytics and symptom criticality tool development