OncoHelper AI: Neural Target Tracking for Cancer in Motion
AI system that treats cancer as a moving target and time as a first-class variable.
11/04/2025. We’re proud to launch the official OncoHelper AI website — a place to discover our mission, follow our progress, and support the fight against cancer. Here you’ll find our projects, updates, and real stories behind the work we do. We’re just beginning — and every supporter helps move this mission forward.
Starting this week, we will be posting weekly updates on this project and its progress. Each update will highlight what has changed in the project sections, the steps we have taken, and what comes next, so you can follow the work in real time. Today we have refreshed several parts of the project to reflect its current status. We are committed to keeping this weekly update cycle, so that our donors and partners always have a clear and confident view of how the project is moving forward.
Update: Starting February 1, 2026, we will publish one project update every two weeks.
Today we are restarting our update cycle for this project.
Since our last update on November 16, 2025, we deliberately paused public postings for two reasons.
First, we focused on completing the technical specification of the Interceptor engine and its supporting modules, so we could build against a stable architecture rather than change direction midstream.
Second, we focused on the data readiness and privacy foundations required for real pediatric clinical data work, including clear rules for how data will be prepared, handled, and validated across partner settings.
Current status: the core system has been designed and specified. Next, we plan to implement and train the first working prototypes using synthetic and retrospective pediatric oncology datasets, then move into formal algorithm validation and co-design of clinician-first interfaces with pediatric oncology teams.
Starting this week, we will publish one update every two weeks. Each update will include what changed on the project page, what we completed during the past two weeks, and what comes next, so donors and partners can follow progress in real time.
Development pace depends on available funding. Support will be used to accelerate implementation, validation, and pilot preparation.
Problem
Pediatric oncology still relies heavily on static snapshots such as imaging, lab values, and periodic clinic visits. These isolated data points often arrive late and do not capture how quickly a child’s physiology, tumor biology, and treatment response can change over time.
Many existing guidelines and AI tools treat patient data as fixed input to be classified, rather than a time-dependent process that must be tracked, interpreted, and anticipated across weeks and months.
This mismatch between how we measure and how the disease behaves contributes to delayed interventions, avoidable toxicity, and missed opportunities to adjust care at the right time.
Solution
This project turns pediatric cancer care into a tracked and time-aware process. OncoHelper AI ingests multi-modal clinical data from hospitals, models how a child’s tumor and treatment response evolve over time, and updates its forecasts as new information becomes available.
Interceptor, our tumor-in-motion tracking and prediction engine currently in development, focuses on what is most likely to happen next and when meaningful change is likely to occur, rather than relying on static snapshots. It surfaces clear, clinician-friendly signals that help care teams decide when to intensify, de-escalate, or change therapy.
In practice, this supports earlier warning of relapse or complications, better timing of key decisions, and more days at home rather than in hospital for children and their families.
In pediatric oncology, care teams face fast-changing conditions. Children are growing, tumors evolve under therapy, and toxicity accumulates over time. Yet much of today’s evidence still comes from static snapshots such as imaging and lab tests that lag behind biological reality.
Most current AI tools treat data as if it were fixed, while real-world oncology is an unfolding process that must be tracked over time and anticipated, not just labeled.
Tracking a moving target, protecting a child
A tumor behaves like a moving target. It can be intercepted only if we detect change early, predict what comes next, and respond in time.
Early detection – the sensing layer
Non-invasive screening of saliva, blood, imaging, and genomics can reveal risk signals long before symptoms appear. Interceptor acts like a high-level sensing and monitoring layer, scanning biological signals to spot emerging patterns early.
Midcourse interception – modeling trajectories
If the tumor is already active, Interceptor switches mode and models how it evolves over time. It tracks how the tumor changes under treatment and forecasts what is most likely to happen next within clinically meaningful time windows. The system reports calibrated uncertainty and flags out-of-scope cases to support safe clinical use.
Persistent monitoring – adaptive updating
When the clinical course changes, Interceptor updates its internal models after new scans, lab results, and treatment cycles. Over time, it builds a patient-specific picture that tracks cancer as an unfolding process rather than a static object.
Interception intelligence
The architecture is designed to combine continuous monitoring, time-based trajectory modeling, and clinician-first signals about when change is likely to occur and what information may be needed next. The difference is that the setting is pediatric care, and the goal is to help teams act earlier, more precisely, and more safely.
This program focuses on tumor-in-motion tracking and time-based decision support for pediatric oncology. It brings together longitudinal AI modeling, multi-modal clinical data, and clinician-first interfaces in a single operational platform.
The scope includes multi-center retrospective validation, federated-ready collaboration where needed, and integration with standard hospital systems. The project will be tested across several pediatric tumor types, with a staged path toward pilot readiness and clinician-controlled use.
The broader aim is to improve outcomes and quality of life for young patients by giving care teams a tool that continuously learns from new information, highlights when meaningful change is likely to occur, and supports earlier, safer intervention decisions.
Planned budget: 350,000 dollars over 24 months for the initial build, validation, and pilot preparation phase.
Earlier detection of relapse, progression, or metastasis signals in pediatric patients.
Fewer treatment delays through clearer time-based forecasting of tumor behavior and treatment response, with risk estimates used as a secondary derived indicator when appropriate.
Improved coordination and shared learning across participating pediatric centers through standardized validation and reporting.
Better toxicity monitoring and more days spent at home rather than in hospital for children and their families.
Clinical and economic evidence to support a compliant pilot pathway and future scale-up.
Core engine: Interceptor, our tumor-in-motion tracking and prediction engine, currently in development. It is designed to model how a child’s tumor and treatment response evolve over time and to forecast what is most likely to happen next as new clinical data becomes available. Risk scoring may be added as a secondary derived output.
Supporting modules:
– Tactician, a scenario module that supports clinician planning by exploring timing-sensitive options and likely consequences.
– Sentry, a continuous surveillance layer that watches for early signals that a patient’s course is deviating from expectations, tuned to reduce alert fatigue.
– A meta-intelligence layer that reconciles signals across data sources, tracks confidence, and enforces safety logic.
Security and data handling: privacy-first data workflows with clear rules for preparation, access control, and auditability. The design is federated-ready so partner sites can participate without moving raw data when needed.
Integration: connection to hospital systems using standard interfaces and formats, with a clinician-first timeline view designed for pediatric oncology workflows.
The core architecture of OncoHelper AI, including the Interceptor engine and its supporting modules, has been designed and specified. We have defined how the system will ingest multi-modal clinical data, model tumor and treatment-response trajectories over time, and connect to hospital systems.
The next phase is to implement and train the first working prototypes using synthetic and retrospective pediatric oncology datasets from partner centers. This stage will benchmark alternative modeling backbones and confirm robust uncertainty reporting and safe behavior on out-of-scope cases.
After that, we will move to formal algorithm validation and co-design of clinician-first interfaces together with pediatric oncologists and nurses, followed by a staged pilot path starting in silent mode before any clinician-controlled recommendation use.
We will publish one progress update every two weeks in the Latest News section so partners and donors can follow the work in real time.
We welcome hospitals, research teams, and donors who share our goal of making pediatric cancer care more time-aware, adaptive, and clinically usable.
Partnerships can support data access for retrospective validation, clinician-led interface co-design, and a staged pilot pathway that begins in silent mode and advances only under clinician control.
Development pace depends on available funding. Support will be applied directly to implementation, validation, and pilot preparation, with progress reported through regular project updates.
Together, we can build a future in which no child’s tumor can outrun our ability to detect change early, anticipate what comes next, and respond in time.
