Building a Natural Language Processing Augmented Information Support System to Enhance Supportive Care for Patients With Prostate Cancer and Families: User-Centered, Iterative Approach.

Background: Patients with prostate cancer and their families face significant challenges during transitions from diagnosis to treatment and posttreatment self-management, particularly in accessing, understanding, and using complex health information. Objective: We aimed to develop the Interactive Prostate Cancer Information, Communication, and Support Program (iPICS), a natural language processing (NLP)-augmented eHealth platform designed to enhance care continuity, support decision-making, and improve health outcomes for patients and families. Methods: This study used an iterative, user-centered design approach to design, develop, and refine iPICS, guided by responsible AI principles. The iterative development process advanced from an initial needs assessment through iterative formative and summative prototype evaluations, culminating in a final evaluation of field deployment readiness via semistructured interviews and focus groups with patients with prostate cancer and family members from diverse sociodemographic backgrounds. Thematic analysis was conducted to identify critical functionalities and content, using double coding and team-based consensus procedures. Participants' feedback was integrated during the platform's refinement to ensure iPICS' usability, accessibility, security, and functionality. Results: A total of 18 patients with prostate cancer and 7 family members participated in 2 semistructured interviews and 17 focus groups. Most participants were older adults and had at least a high school education. Participants identified 5 major themes relevant to iPICS development: functional requirements, user interface design recommendations, content and visualization needs, program delivery preferences, and privacy and data security concerns. These themes informed the iPICS prototype design, development, and refinement, which include 3 core components: Inform, a multimedia health information resource hub; Dialog, an NLP-augmented consultation recording summarization tool for patient-provider communication; and Snap, a moderated online peer-support forum. Key features of iPICS include: Inform is an evidence-based, guideline-informed multimedia health information paired with National Institutes of Health-sponsored MedlinePlus papers; Dialog is an NLP-powered recording with keyword extraction and hyperlinking; and Snap is peer-support functionalities moderated by nurses to ensure safety and reliability. iPICS development was also compliant with HIPAA (Health Insurance Portability and Accountability Act) standards and aligned with user needs while ensuring usability throughout deployment. Conclusions: The NLP-augmented iPICS was developed using an iterative user-centered design approach to support patients with prostate cancer and their families. It offers a scalable solution for ethical, transparent, and inclusive supportive survivorship care, particularly during critical care transitions. As a formative qualitative study with a small sample, this phase did not evaluate patient- or family-reported outcomes. Our ongoing studies will evaluate the effects of iPICS on patient- and family-reported outcomes and explore adaptation to other cancers and chronic conditions.