New Mathematical Challenges Introduced in ROS Cycle #11
Description
New Mathematical Challenges Introduced in ROS Cycle
This document summarizes the new and explicit mathematical problems that the current ROS Cycle implementation is now attempting to solve. These are not bugs, but formal challenges that emerge once the system moves from simple date arithmetic to probabilistic modeling.
1. Translating Biological Priors into Computable Models
Problem
Human menstrual biology provides approximate priors (e.g., ovulation ≈ 14 days before next period), but these priors are not strict laws.
Challenge
How do we encode biological assumptions into equations without claiming medical certainty?
Mathematical Question
How should fixed biological priors be represented: as constants, soft centers, or distributions?
2. Modeling Ovulation as a Continuous Likelihood Curve
Problem
Ovulation is not directly observable in this system, yet the model must represent its likelihood across days.
Challenge
The system introduces a Gaussian-shaped curve to represent ovulation likelihood across the cycle.
Mathematical Question
Is ovulation better represented as:
- a relative likelihood score, or
- a normalized probability distribution over the cycle?
3. Defining Fertile Windows from Continuous Signals
Problem
Users expect discrete labels (“low”, “medium”, “high fertility”), but the model produces continuous values.
Challenge
Choosing where to draw boundaries on a continuous curve.
Mathematical Question
Should fertile windows be defined by:
- fixed thresholds,
- percentiles of likelihood mass, or
- cumulative probability regions?
4. Quantifying Cycle Variability and Uncertainty
Problem
Cycle variability affects prediction reliability, but is not directly observable beyond user input.
Challenge
Translating variability (± days) into uncertainty within predictions.
Mathematical Question
How should variability reshape:
- curve width,
- fertile window size, and
- confidence scores?
5. Designing a Meaningful Confidence Metric
Problem
Users need to know how much to trust predictions.
Challenge
Current confidence is linearly derived from variability, which may not reflect real uncertainty.
Mathematical Question
What mathematical form best represents trust in predictions given incomplete data?
6. Integrating Symptom-Based Risk Signals (PCOD)
Problem
PCOD symptoms imply irregular cycles, but are not definitive diagnoses.
Challenge
Incorporating qualitative risk signals without medical overreach.
Mathematical Question
How should soft risk indicators alter probabilistic models without becoming diagnostic?
7. Separating Determinism from Probabilistic Inference
Problem
Some outputs (calendar dates) are deterministic; others (fertility, ovulation) are probabilistic.
Challenge
Preventing users from interpreting probabilistic outputs as certainties.
Mathematical Question
How do we clearly distinguish deterministic computation from probabilistic inference in both math and UI?
8. Maintaining Internal Consistency Across the Model
Problem
As more signals are added, inconsistencies can arise between equations, labels, and UI claims.
Challenge
Ensuring that every displayed value is mathematically defensible.
Mathematical Question
What invariants must always hold true across the model (e.g., bounded outputs, monotonic confidence decay)?
Summary
ROS Cycle is no longer just predicting dates.
It is now tackling the broader problem of how to responsibly model human biological uncertainty using minimal, local-only data.
These challenges define the future mathematical direction of the project and should guide all future changes to the model.