For Families.

The thing I wished existed

I built this for personal reasons; the short version is on the motivation page. What I wanted was a useful sense of the likely trajectory: something more specific than “this disease is different for everyone,” but less certain than the disease allows. It would show a range, explain why that range is wide, and arrive early enough to help a family prepare.

The long-term goal is to provide that kind of information in plain language, with disease-specific forecasts that explain what is likely, what remains unknown, and how the uncertainty affects real decisions.

Why it doesn't exist yet

No forecast should be offered to families until the evidence supports it. Poorly calibrated guidance can do real harm. Two conditions must be met:

1. Evidence beyond one cohort. PRO-ACT can support the development and internal evaluation of an ALS model, but a clinical-trial cohort is not representative enough to justify a family-facing tool. That would require external validation in representative data and an appropriate clinical partnership. Longitudinal speech and Alzheimer’s cohorts remain separately governed and are not guaranteed on any timeline.
2. Reliable calibration. The model must produce uncertainty estimates I would trust for my own family: how quickly function may change, which functions may change first, and how wide the plausible range is.

The molecular and HRS studies establish methods for that work, but neither produces a family-facing forecast. Nothing will be offered to families until both conditions are met.

If you're a family member

The build log is what I can offer right now. It documents what the models can and cannot do, in plain language, as the work develops.

If you want to talk, be notified if this becomes ready, or hold the work to its stated standard, reach out directly. Families who have been through this deserve a clear update even if the necessary data never becomes available or the model fails to meet the standard.