Previously submitted to: JMIR Cancer (no longer under consideration since Jun 11, 2026)
Date Submitted: Jun 1, 2026
Open Peer Review Period: Jun 11, 2026 - Jun 11, 2026
(closed for review but you can still tweet)
NOTE: This is an unreviewed Preprint
Warning: This is a unreviewed preprint (What is a preprint?). Readers are warned that the document has not been peer-reviewed by expert/patient reviewers or an academic editor, may contain misleading claims, and is likely to undergo changes before final publication, if accepted, or may have been rejected/withdrawn (a note "no longer under consideration" will appear above).
Peer review me: Readers with interest and expertise are encouraged to sign up as peer-reviewer, if the paper is within an open peer-review period (in this case, a "Peer Review Me" button to sign up as reviewer is displayed above). All preprints currently open for review are listed here. Outside of the formal open peer-review period we encourage you to tweet about the preprint.
Citation: Please cite this preprint only for review purposes or for grant applications and CVs (if you are the author).
Final version: If our system detects a final peer-reviewed "version of record" (VoR) published in any journal, a link to that VoR will appear below. Readers are then encourage to cite the VoR instead of this preprint.
Settings: If you are the author, you can login and change the preprint display settings, but the preprint URL/DOI is supposed to be stable and citable, so it should not be removed once posted.
Submit: To post your own preprint, simply submit to any JMIR journal, and choose the appropriate settings to expose your submitted version as preprint.
Warning: This is an author submission that is not peer-reviewed or edited. Preprints - unless they show as "accepted" - should not be relied on to guide clinical practice or health-related behavior and should not be reported in news media as established information.
AI-Enabled Mobile Self-Examination and Early Detection Centers (CADI) for Breast Cancer in Vulnerable Communities in Colombia: Protocol for a Mixed Methods Feasibility and Implementation Study
ABSTRACT
Background:
Abstract
Background:
Breast cancer remains a major public health challenge in Colombia and the Region of the Americas, with inequities in timely detection and referral disproportionately affecting rural, low-income, and geographically isolated populations. The Centers for Self-Examination and Intelligent Detection (Centros de Autoexploración y Detección Inteligente; CADI) project proposes a mobile, solar-powered, AI-enabled unit that combines guided breast self-examination education, infrared thermal imaging as an adjunctive triage tool, and interoperable referral alerts to health insurers and care networks.
Objective:
This protocol describes the design and planned mixed methods feasibility evaluation of CADI for early breast cancer detection support in vulnerable communities in Cartagena and selected rural Colombian settings.
Methods:
The study will use a sequential mixed methods feasibility and implementation design informed by Challenge-Based Learning. Phase 1 will refine user, clinical, and institutional requirements through epidemiologic review, stakeholder mapping, and interviews with community representatives and health professionals. Phase 2 will specify the CADI prototype, including privacy-preserving data capture, AI-assisted thermal-image risk classification, guided education, SMS or call-based alerts, and HL7/FHIR-compatible transfer to health entities when connectivity is available. Phase 3 will pilot the model in prioritized community settings. Feasibility outcomes will include reach, uptake, completion of guided sessions, system uptime, successful synchronization, referral activation, user acceptability, and approximate unit cost per completed encounter. Quantitative data will be summarized descriptively and, where appropriate, compared across sites; qualitative data will be analyzed thematically.
Results:
The ABR design phase produced a mobile CADI model targeting adults in vulnerable communities, a multichannel implementation strategy, a stakeholder partnership map, and preliminary cost estimates. Operational assumptions include approximately 300 users per cabin per month, monthly operating costs of COP $7.0 million to COP $9.5 million, an estimated cost of COP $23,000 to COP $31,000 per user, and a prototype creation cost of COP $27.8 million to COP $44.2 million per cabin. No participant-level diagnostic accuracy or effectiveness outcomes have yet been collected.
Conclusions:
CADI is a context-adapted digital health and community outreach proposal intended to reduce access barriers to breast cancer early detection pathways. Its scientific value will depend on ethical pilot testing, clinical validation against accepted diagnostic standards, careful communication that thermography is adjunctive and not a replacement for mammography, and robust governance for privacy, interoperability, and referral continuity. Keywords: Breast Neoplasms; Early Detection of Cancer; Artificial Intelligence; Thermography; Mobile Health Units; Telemedicine; Health Equity; Vulnerable Populations; Colombia; Feasibility Studies
Objective:
Objective The objective of this manuscript is to describe the CADI intervention and the protocol for a mixed methods feasibility and implementation study evaluating its acceptability, operational performance, referral functionality, and readiness for clinical validation in vulnerable communities in Cartagena and selected rural Colombian settings.
Methods:
Methods Study Design This is a protocol for a sequential mixed methods feasibility and implementation study. The design adapts the original Challenge-Based Learning process into a publishable research protocol with four connected stages: problem identification, contextual needs assessment, CADI prototype specification, and pilot feasibility evaluation. The quantitative component will estimate reach, uptake, system performance, referral activation, and operational costs. The qualitative component will explore barriers, acceptability, perceived trust, usability, cultural adaptation, privacy concerns, and implementation feasibility from the perspectives of community members, health professionals, institutional decision makers, and implementation staff. Setting The proposed setting includes vulnerable urban neighborhoods in Cartagena de Indias, Bolívar, Colombia, and selected rural or peri-urban communities with limited access to specialized breast imaging and oncology referral pathways. Candidate locations include community centers, public spaces, transportation nodes, health posts, schools, and sites prioritized by local health secretariats or health insurers based on epidemiologic indicators and access barriers. Target Population and Eligibility The source ABR document identifies the target population as adults older than 20 years without a previous breast cancer diagnosis who live in communities with limited access to health services, along with people in accessible public spaces who may still face informational, economic, or time-related barriers to preventive screening. For pilot evaluation, the primary target group will be adults aged 30 to 70 years, prioritizing women and including men when clinically relevant, with no known current breast cancer diagnosis. Exclusion criteria will include current breast cancer treatment, inability to provide informed consent, medical instability requiring urgent care, or refusal of privacy and data-use terms. CADI Intervention Components Component Description JMIR-relevant implementation issue Mobile/modular cabin Compact, accessible, private unit with ramp access, thermal protection, LED lighting, solar panels, and backup battery. Feasibility depends on site permissions, climate resilience, maintenance, and user privacy. Guided education Interactive touch screen and culturally adapted tutorial for breast self-examination and prevention education. Language, health literacy, and culturally safe communication must be validated with the target community. AI-assisted thermal imaging Infrared thermal camera captures temperature-pattern data for risk triage and flags suspicious findings. Must be communicated as adjunctive triage, not as diagnostic replacement for mammography or medical evaluation. Data system and interoperability Local encrypted storage, user code, delayed synchronization when connectivity is intermittent, and HL7/FHIR-compatible transfer to EPS or health-system platforms. Requires cybersecurity, consent, data minimization, audit logs, and interoperability testing. Referral and follow-up alerts SMS, telephone, email, software alerts, or community health worker contact for suspicious findings and appointment navigation. Success depends on confirmed referral completion and continuity from triage to diagnostic confirmation. Community engagement Training of community health agents, educational campaigns, and partnerships with local leaders. Trust, retention, and equity depend on co-design and bidirectional feedback. Study Phases Phase Purpose Core activities Outputs 1. Needs and context assessment Define barriers, population needs, and institutional requirements. Review public epidemiologic data; map routes of care; interview health workers, community leaders, and potential users; identify language and accessibility needs. Implementation requirements, stakeholder map, eligibility refinement. 2. Prototype specification Translate requirements into the CADI operating model. Define cabin design, workflow, data fields, privacy safeguards, AI triage process, referral triggers, and connectivity process. Prototype specification, standard operating procedures, data dictionary. 3. Pilot feasibility testing Assess reach, usability, system function, referral activation, and cost in selected sites. Deploy CADI; collect encounter data; monitor uptime; administer acceptability survey; document referrals; conduct interviews. Feasibility indicators, acceptability metrics, implementation barriers. 4. Optimization and scale-up planning Refine CADI before broader clinical validation. Integrate quantitative and qualitative findings; revise workflow; estimate budget; define larger validation study. Optimized model, scale-up plan, future validation protocol.
Results:
Quantitative Outcomes • Reach: number of community members approached, eligible, enrolled, and completing a CADI session. • Uptake: proportion of eligible individuals who consent and complete guided education and triage. • Operational feasibility: cabin uptime, number of technical interruptions, successful local capture, successful synchronization, and time from encounter to data transfer. • Referral activation: number and proportion of suspicious or moderate/high-risk flags that generate a documented referral alert. • Referral continuity: documented completion of follow-up appointment or confirmatory imaging when data-sharing agreements permit tracking. • Cost: monthly operating cost, cost per completed CADI encounter, and estimated cost per referral activated. • User-reported outcomes: perceived privacy, comprehension of guidance, satisfaction, trust, and willingness to recommend the service. Qualitative Outcomes Semi-structured interviews or focus groups will explore acceptability, perceived usefulness, cultural fit, barriers to participation, concerns about AI-based risk triage, perceptions of privacy, referral navigation challenges, and stakeholder recommendations for scale-up. Participants may include CADI users, community health agents, nursing staff, EPS representatives, physicians, local health authority representatives, and implementation personnel. Data Collection Data sources will include structured encounter records, system logs, user surveys, referral records, cost logs, public epidemiologic indicators, and qualitative interviews. Encounter records will avoid unnecessary identifiers and will use a unique study code. Any personally identifying information needed for referral will be stored separately from research data and transferred only under an approved consent process and data-sharing agreement. Analysis Quantitative feasibility outcomes will be summarized using counts, percentages, medians, interquartile ranges, means, and SDs as appropriate. Where sample size and data quality permit, comparisons across sites will use chi-square tests or Fisher exact tests for categorical variables and t tests or nonparametric alternatives for continuous variables. Multivariable exploratory models may examine associations among demographic factors, site characteristics, uptake, and referral completion. Qualitative data will be analyzed using reflexive thematic analysis. Mixed methods integration will occur through a joint display linking quantitative feasibility indicators with qualitative explanations and implementation recommendations. Data Security and Privacy The CADI data architecture will use privacy by design principles, including data minimization, local encryption, role-based access, separation of referral identifiers from research data, audit logs, and secure synchronization. The source ABR design proposes AES-256-type encryption, delayed synchronization when internet connectivity is unavailable, and interoperability through HL7 or FHIR standards. The protocol will operationalize these requirements in a data management plan approved by the ethics committee and institutional partners. Ethical Considerations The current manuscript reports a design and protocol adaptation and does not report participant-level data. Before any pilot involving human participants, the study team will obtain ethics committee approval from Universidad del Sinú or the competent institutional review body and, when required, authorization from participating health care institutions and community sites. Written or electronic informed consent will be obtained before participation. The consent process will explain that CADI is not a definitive diagnostic test and does not replace mammography, ultrasound, biopsy, or medical evaluation. Participants with suspicious findings will receive referral instructions and navigation support. Additional safeguards will be used for persons with low literacy, limited digital access, disability, or social vulnerability. Results Study Status and ABR Design Outputs The original ABR process generated a publishable intervention concept but did not generate participant-level clinical, diagnostic accuracy, or effectiveness data. The results below therefore report design outputs, implementation specifications, and planned evaluation indicators. This distinction is essential for JMIR eligibility and prevents overstatement of the maturity of the evidence. Domain Design output derived from ABR document Planned feasibility indicator Public health problem Breast cancer mortality and access inequity in Cartagena and vulnerable Colombian communities. Community reach and proportion of participants overdue or disconnected from screening pathways. Intervention model CADI mobile/modular unit with guided self-examination, AI-assisted thermal imaging, touch-screen education, solar power, local storage, and referral alerts. Session completion rate, user satisfaction, system uptime, and alert generation success. Population Adults in vulnerable communities, prioritizing women aged 30-70 years and including men when clinically relevant. Enrollment, refusal rate, age distribution, socioeconomic proxy indicators, and referral needs. Implementation partners EPS, health secretariats, universities, technology suppliers, telecom operators, community organizations, NGOs, and financing partners. Number of active agreements, referral pathway readiness, and response time. Cost model Monthly operating cost per cabin estimated at COP $7.0-$9.5 million; approximate cost per user COP $23,000-$31,000 assuming 300 users per month; prototype creation cost COP $27.8-$44.2 million. Observed cost per completed encounter and sensitivity analysis by volume. Solution Evaluation Findings Reorganized for JMIR The ABR evaluation compared local, national, and international approaches. Community education and campaigns were judged useful for awareness but limited by continuity and financing. Hospital-based strategies improve diagnostic capacity but may not resolve rural access barriers. University and NGO partnerships strengthen education and research translation but require sustained implementation structures. National policy frameworks enable access and referral but depend on infrastructure. International mobile screening models demonstrate the value of mobility, but sustainability, equipment maintenance, and contextual adaptation remain critical limitations. CADI is designed to integrate these lessons by combining community education, mobile deployment, interoperable referral, and a costed implementation model. Preliminary Business and Sustainability Outputs The source ABR Canvas proposes a hybrid implementation model. Institutional customers include EPS, hospitals, clinics, municipal and departmental health secretariats, and public health programs. A community-facing model may include subsidized or low-cost access in public venues through grants, public contracts, corporate social responsibility funding, or health insurer partnerships. This business logic is not presented as evidence of economic effectiveness; rather, it is a sustainability hypothesis to be tested during implementation research.
Conclusions:
Discussion Principal Considerations CADI addresses a plausible implementation gap: people in vulnerable communities often need more than diagnostic technology; they need education, culturally appropriate contact, privacy, navigation, and reliable linkage to confirmatory care. The CADI concept is strongest when framed as a community-based digital health and referral-navigation platform rather than as a standalone diagnostic device. This framing aligns the intervention with public health goals, Colombian cancer-control legislation, and ethical communication requirements for adjunctive technologies. Comparison With Prior Approaches Mobile mammography and outreach models have demonstrated the value of taking services closer to communities, while digital health interventions can support navigation, reminders, and continuity. CADI differs from conventional mobile screening by emphasizing guided self-examination education, AI-supported triage, solar-powered deployment, and interoperability with EPS systems. However, the scientific evidence for infrared thermography remains mixed and evolving. Recent reviews suggest renewed interest because of improved imaging and machine learning, but regulatory guidance remains clear that thermography must not replace mammography.8,9 CADI should therefore be evaluated as an adjunctive access and triage innovation, with confirmatory pathways built into the protocol. Policy and Health Equity Implications The Colombian policy environment supports prevention, early detection, timely diagnosis, rehabilitation, and comprehensive cancer care.4-7 CADI may operationalize these policy goals if it improves access in communities where screening infrastructure is limited, if it links people with suspicious findings to existing clinical pathways, and if it generates actionable data for health insurers and public health authorities. The equity promise of CADI will depend on avoiding algorithmic bias, minimizing out-of-pocket payments for low-income users, providing multilingual and low-literacy materials, and ensuring that no participant is left with a risk flag without navigation to confirmatory care. Limitations • The current ABR document is a proposal and design exercise; it does not contain clinical validation, diagnostic accuracy estimates, participant outcomes, or health economic evaluation. • Thermal imaging may generate false reassurance or false alarms if communicated poorly; CADI must be described as adjunctive triage and education, not as a replacement for standard screening or diagnostic imaging. • Interoperability with EPS systems will require legal agreements, technical integration, privacy safeguards, and workflow alignment that may vary by institution. • Rural deployment may face electricity, connectivity, transportation, security, maintenance, and staffing constraints. • AI performance may vary by age, breast density, skin tone, ambient temperature, comorbidities, imaging protocol, and training data representativeness. • The cost assumptions are preliminary and should be validated using real procurement quotes, maintenance records, depreciation, staff costs, and site-specific volume estimates. Future Research The next step should be an ethics-approved feasibility pilot followed by a clinical validation study comparing CADI risk classification and referral performance with accepted diagnostic pathways such as mammography, ultrasound, clinical breast examination, and biopsy when indicated. Future studies should report diagnostic accuracy, equity outcomes, time to diagnosis, referral completion, user acceptability, cost-effectiveness, and implementation determinants using recognized reporting guidelines. Conclusions CADI is a promising community-oriented digital health protocol for early breast cancer detection support in vulnerable Colombian settings. Its publishable contribution is strongest as a JMIR Research Protocols submission that transparently describes the intervention, planned mixed methods evaluation, ethical safeguards, and implementation hypotheses. Publication as an Original Paper in JMIR Cancer or JMIR Formative Research should be pursued after pilot data are collected and analyzed. Clinical Trial: References 1. World Health Organization. Breast cancer. Published April 16, 2026. Accessed June 1, 2026. https://www.who.int/news-room/fact-sheets/detail/breast-cancer 2. Pan American Health Organization. Breast cancer. Accessed June 1, 2026. https://www.paho.org/en/topics/breast-cancer 3. Pan American Health Organization. Cancer. Accessed June 1, 2026. https://www.paho.org/en/topics/cancer 4. Congreso de Colombia. Ley 1384 de 2010: Ley Sandra Ceballos, por la cual se establecen las acciones para la atención integral del cáncer en Colombia. Accessed June 1, 2026. https://www.funcionpublica.gov.co/eva/gestornormativo/norma.php?i=39368 5. Congreso de Colombia. Ley 2194 de 2022: por medio de la cual se modifica la Ley 1384 de 2010. Accessed June 1, 2026. https://www.alcaldiabogota.gov.co/sisjur/normas/Norma1.jsp?i=120363 6. Congreso de Colombia. Ley 2360 de 2024: por medio de la cual se modifica la Ley 1384 de 2010. Accessed June 1, 2026. https://www.funcionpublica.gov.co/eva/gestornormativo/norma.php?i=243296 7. Ministerio de Salud y Protección Social. Resolución No. 3280 de 2018. Published August 2, 2018. Accessed June 1, 2026. https://www.minsalud.gov.co/Normatividad_Nuevo/Resoluci%C3%B3n%20No.%203280%20de%2020183280.pdf 8. US Food and Drug Administration. Breast cancer screening: thermogram no substitute for mammogram. Published October 6, 2023. Accessed June 1, 2026. https://www.fda.gov/consumers/consumer-updates/breast-cancer-screening-thermogram-no-substitute-mammogram 9. Goñi-Arana A, Pérez-Martín J, Díez FJ. Breast thermography: a systematic review and meta-analysis. Syst Rev. 2024;13:318. doi:10.1186/s13643-024-02708-9 10. Mashekova A, et al. Review of artificial intelligence techniques for breast cancer detection using medical imaging modalities. Accessed June 1, 2026. https://pmc.ncbi.nlm.nih.gov/articles/PMC12561872/ 11. Health Level Seven International. HL7 FHIR Specification. Accessed June 1, 2026. https://hl7.org/fhir/ 12. National Institute of Standards and Technology. Advanced Encryption Standard (AES), FIPS 197-upd1. Published 2023. Accessed June 1, 2026. https://www.nist.gov/publications/advanced-encryption-standard-aes-0 13. Departamento Administrativo Nacional de Estadística. Estadísticas vitales: nacimientos y defunciones. Accessed June 1, 2026. https://www.dane.gov.co/index.php/estadisticas-por-tema?id=34 14. Secretaría de Planeación Distrital de Cartagena. Mortalidad cáncer de mama: histórico tasa por 100,000 mujeres. Accessed June 1, 2026. https://planeacion.cartagena.gov.co/sites/default/files/2024-10/CANCER%20DE%20MAMA-MORTALIDAD-HISTORICO-TASA%20X%20100.000%20HAB.xlsx 15. JMIR Publications. Instructions for Authors. Accessed June 1, 2026. https://www.jmir.org/author-information/instructions-for-authors 16. JMIR Publications. Submitting Your Manuscript to JMIR Publications: A Guide for Authors. Updated May 11, 2026. Accessed June 1, 2026. https://support.jmir.org/hc/en-us/articles/37982552280987-Submitting-Your-Manuscript-to-JMIR-Publications-A-Guide-for-Authors 17. JMIR Publications. What are JMIR's guidelines for article titles? Updated February 18, 2025. Accessed June 1, 2026. https://support.jmir.org/hc/en-us/articles/115002943791-What-are-JMIR-s-guidelines-for-article-titles 18. JMIR Research Protocols. Focus and Scope. Accessed June 1, 2026. https://www.researchprotocols.org/about-journal/focus-and-scope
Citation
Request queued. Please wait while the file is being generated. It may take some time.
Copyright
© The authors. All rights reserved. This is a privileged document currently under peer-review/community review (or an accepted/rejected manuscript). Authors have provided JMIR Publications with an exclusive license to publish this preprint on it's website for review and ahead-of-print citation purposes only. While the final peer-reviewed paper may be licensed under a cc-by license on publication, at this stage authors and publisher expressively prohibit redistribution of this draft paper other than for review purposes.