Artificial Intelligence Detection of National Clinical Practice Gaps in Hormone-Sensitive Prostate Cancer (HSPC) in Brazil: A Patient-Centered Framework
Author(s)
diana mk tan, BA1, Mingxuan Lim, Bachelor of Science2.
1Advocates For Advancing Health, SINGAPORE, Singapore, 2Advocates For Advancing Health, Singapore, Singapore.
1Advocates For Advancing Health, SINGAPORE, Singapore, 2Advocates For Advancing Health, Singapore, Singapore.
OBJECTIVES: Guidelines endorse early androgen receptor pathway inhibitor intensification in HSPC, yet uptake remains suboptimal in Brazil, impacting timely patient access to appropriate care. Current gap identification methods lack scalability and precision targeting. We developed an AI based digital behavior framework to quantify specialty specific clinical practice gaps and inform value based access strategies.
METHODS: The validated AI algorithm (Sqreem Technologies) analyzed >15 billion anonymized digital actions over a period of approximately 90 days, representing >85% of all online activity in Brazil’s HSPC space. Relevant online activity by Urologists (UROs) and Oncologists (ONCs) was identified. HCP identification used a probabilistic model, validated by independent third-party audits (88-96% accuracy). The AI model identified HCP questions in specific topics and measured the intersection with available resources. Unmet interests (whitespace, WS) were flagged as educational gaps. Nonmetastatic (nm)HSPC and mHSPC were modelled separately. Data were fully anonymized and GDPR-compliant.
RESULTS: A total of 543,111 WS signals were detected. WS were marginally higher in mHSPC (UROs 14%, ONCs 12%) than nmHSPC (UROs 12%, ONCs 11%).
Patient-related value and access WS dominated, representing 81% of total WS: diagnosis, risk stratification (UROs 49%, ONCs 36%); treatment selection, intensification, sequencing and deintensification (URO 19%, ONCs 26%); patient education, AE, QoL management (URO 13%, ONCs 18%). The significant overlap between specialties highlighted opportunities for multidisciplinary interventions.
CONCLUSIONS: This AI approach enables near real-time detection of national clinical practice gaps, overcoming limitations of traditional, sample-based methods. The methodology is scalable and adaptable across tumor types and health systems.
By generating patient-centered evidence at scale, this approach identified persistent gaps and informed the development of educational and decision support tools to optimize shared stakeholder decision making, improve patient outcomes, and accelerate equitable access to guideline-concordant care.
METHODS: The validated AI algorithm (Sqreem Technologies) analyzed >15 billion anonymized digital actions over a period of approximately 90 days, representing >85% of all online activity in Brazil’s HSPC space. Relevant online activity by Urologists (UROs) and Oncologists (ONCs) was identified. HCP identification used a probabilistic model, validated by independent third-party audits (88-96% accuracy). The AI model identified HCP questions in specific topics and measured the intersection with available resources. Unmet interests (whitespace, WS) were flagged as educational gaps. Nonmetastatic (nm)HSPC and mHSPC were modelled separately. Data were fully anonymized and GDPR-compliant.
RESULTS: A total of 543,111 WS signals were detected. WS were marginally higher in mHSPC (UROs 14%, ONCs 12%) than nmHSPC (UROs 12%, ONCs 11%).
Patient-related value and access WS dominated, representing 81% of total WS: diagnosis, risk stratification (UROs 49%, ONCs 36%); treatment selection, intensification, sequencing and deintensification (URO 19%, ONCs 26%); patient education, AE, QoL management (URO 13%, ONCs 18%). The significant overlap between specialties highlighted opportunities for multidisciplinary interventions.
CONCLUSIONS: This AI approach enables near real-time detection of national clinical practice gaps, overcoming limitations of traditional, sample-based methods. The methodology is scalable and adaptable across tumor types and health systems.
By generating patient-centered evidence at scale, this approach identified persistent gaps and informed the development of educational and decision support tools to optimize shared stakeholder decision making, improve patient outcomes, and accelerate equitable access to guideline-concordant care.
Conference/Value in Health Info
2025-11, ISPOR Europe 2025, Glasgow, Scotland
Value in Health, Volume 28, Issue S2
Code
MSR34
Topic
Methodological & Statistical Research, Patient-Centered Research, Real World Data & Information Systems
Topic Subcategory
Artificial Intelligence, Machine Learning, Predictive Analytics
Disease
Oncology