back to the newsroom
In Rochester, Sofya presented a research direction that connects longitudinal data, disease foundation models and care gaps to reviewable clinical actions.
The first Brazilian startup accepted into Mayo Clinic Platform_Accelerate, Sofya brought to the program's final showcase in Rochester a more demanding question than demonstrating a good model: how to make precision care viable at scale, within the context of each health system? The proposed answer combines automated real-time clinical reasoning with a longitudinal representation able to link risk, care and outcome.
Clinical intelligence only matters when it appears in time to support a decision. That shifts the product's center from the finished note to the still-open course of action: recognizing what changed, gathering the relevant context, making limits explicit and handing the professional a hypothesis that can be reviewed.
The final presentation organized the research as a single flow: represent the journey, create clinical references, measure distances and turn recurring deviations into reviewable gaps.
Risk, care and outcomes are aligned in time before any comparison.
Literature, guidelines, clusters and expert review form clinical anchors.
Each case is compared against the relevant care dimensions, without erasing its singularity.
Recurring patterns make variations and care opportunities visible to the system.
Research direction presented at the Mayo Clinic Platform_Accelerate final showcase; any clinical application requires local validation and professional review.
In the initial prostate cancer scope, the research combines literature, guidelines, data-derived clusters and expert review to build clinical personas. They do not replace the patient. They work as multidimensional references to interpret severity, trajectory and care decisions without reducing the case to a single code or score.
Man in his early 60s with bone pain, low back pain and fatigue, on a high-risk trajectory. HRD testing was ordered and PARP inhibitor use is consistent with an HRR-altered disease pathway or protocol, although the molecular result is not recorded.
Baseline PSA of 45 ng/mL. Digital rectal exam performed, no result recorded. mpMRI and ultrasound-guided biopsy confirmed Gleason 4+3=7 adenocarcinoma, Grade Group 3. The record indicates metastatic disease, stage 4, with PSMA-PET performed (CPT 78811).
Androgen deprivation therapy with an LHRH agonist or antagonist, PARP inhibitor and bone-modifying agent such as denosumab or zoledronic acid. Definitive local treatment was not documented.
PSA 45 → 5 → 25 → 25 ng/mL. Serum testosterone between 150 and 240 ng/dL, above castration level.
The biochemical rise occurs with testosterone above 50 ng/dL, suggesting incomplete castration. The available record does not support criteria for castration-resistant prostate cancer.
By measuring the distance between the patient and these references over time, the model can make recurring deviations visible: a transition that did not happen, late monitoring, a hard-to-explain variation or a care dimension that deserves review. The goal is not to produce an automatic verdict, but to reveal reasons for gaps that a professional and an institution can examine.
The trajectory approaches an anchor, but a test, a response or a step of care may be missing.
There is partial similarity. The order of events, treatment response or follow-up deserves review.
The case moves away from the anchors. There may be missing information, an atypical presentation or a trajectory that needs reassessment.
Synthetic persona to demonstrate clustering and comparison. Map positions, connections and anchors are illustrative and do not represent a validated clinical measure or a therapeutic recommendation.
This architecture connects efficiency and precision. In the consultation, it supports documentation, guidelines and planning. At scale, it can improve cohort resolution, strengthen real-world evidence and help health systems locate avoidable variation. The next research step is to bring this model back into the clinical flow and measure whether the suggested actions produce real impact.
The work presented does not turn a research hypothesis into a clinical claim. Each institution, population and purpose requires its own validation, professional oversight and governance. The advance lies in formulating a verifiable architecture for precision care: longitudinal context, explicit references, interpretable distances and learning with provenance.
The showcase brought together product direction, precision care research and the validation context built over the program.
We use cookies to measure audience and improve the site. Learn more in our Privacy Policy.
