Researchers at Nagoya University have developed a machine learning system that can identify how fast ALS will advance in a patient — and which body functions will decline first — using only information available at the initial medical visit.
Amyotrophic Lateral Sclerosis (ALS) is one of the most unpredictable and devastating neurological diseases known to medicine. Two patients diagnosed on the same day can have completely different outcomes — one may decline slowly over a decade, while another deteriorates rapidly within months. Until now, doctors have had almost no reliable way to tell which path a patient will follow.
That may be about to change.
A team of scientists at Nagoya University in Japan has built an AI-powered tool called DiSPAH — short for Disease progression Speed and PAthway Heterogeneity — that separates two distinct dimensions of ALS progression that previous tools lumped together: how fast the disease advances, and in what order the body’s functions break down.
The findings were published in the peer-reviewed journal npj Digital Medicine.
Why This Matters: ALS Is Not One Disease, It’s Many
ALS is a fatal neurodegenerative condition that progressively destroys the nerve cells controlling movement, speech, and breathing. But researchers have long recognized that “ALS” is not a single uniform disease — it behaves very differently from person to person.
Some patients lose the use of their limbs first. Others lose speech and the ability to swallow. Some decline over years; others over months. This enormous variability has made it nearly impossible to design clinical trials, personalize treatment plans, or give families meaningful guidance about what to expect.
The DiSPAH system was designed specifically to address this challenge by treating speed and pathway as two separate, independent variables — something no previous AI tool had cleanly achieved.
What the AI Found
The Nagoya University team trained DiSPAH on data from 264 patients with limb-onset ALS — the form of the disease where symptoms begin in the arms or legs rather than in speech or swallowing muscles. They then validated their findings against a much larger, independent dataset of 2,565 ALS patients.
The AI identified six distinct patterns of disease progression among the patients. Some patients followed a trajectory of slow motor decline with minimal impact on speech or breathing. Others experienced rapid deterioration across multiple systems simultaneously.
Crucially, the research confirmed that speed and pattern are independent of each other. A patient could follow a severe decline pattern but at a slow pace — or experience a milder pattern but at rapid speed. These two dimensions simply cannot be inferred from each other, which is why separating them matters so much clinically.
Perhaps most significantly, DiSPAH was able to predict a patient’s progression speed and broad decline pattern from information available at the very first medical visit — basic functional assessments combined with the presence of certain genetic mutations.
A Genetic Clue Hidden in the Data
One of the study’s most actionable findings involves a specific gene mutation. Patients carrying a mutation in a gene called C9orf72 — already known to be associated with ALS risk — were found to have significantly faster disease progression than patients without it.
Deeper analysis of motor neurons derived from patient iPS cells (laboratory-grown stem cells that can become any cell type) suggested that disrupted protein production processes and elevated oxidative stress within these cells may be driving faster decline.
This genetic link could eventually allow doctors to flag high-risk patients at diagnosis and prioritize them for aggressive early intervention or clinical trial enrollment.
What This Means for Patients and Families
The practical implications of DiSPAH extend in several directions simultaneously:
For patients and caregivers: Earlier, more accurate prognosis means families can make informed decisions about care planning, home modifications, and support services — without waiting months or years for the disease’s trajectory to become clear on its own.
For clinicians: Doctors can tailor treatment strategies from the outset rather than adopting a watch-and-wait approach. For example, patients predicted to show rapid respiratory decline could be prioritized for early respiratory support monitoring.
For clinical trials: One of the biggest challenges in ALS research is that clinical trials often mix together patients with very different disease trajectories, making it difficult to detect whether a drug is actually working. DiSPAH could allow trial designers to group participants by their progression profile — dramatically improving the statistical power of future ALS drug trials and potentially accelerating the path to approved treatments.
How It Fits Into the Broader AI-in-Medicine Trend
The DiSPAH study is part of a rapidly growing body of research applying machine learning to diseases where outcome variability has long made personalized medicine difficult. Similar AI approaches have recently been applied to Parkinson’s disease progression, multiple sclerosis relapse prediction, and cancer staging.
What distinguishes this particular work is its focus on disentangling two variables that look similar on the surface but are mechanistically distinct — a methodological contribution that may be as significant as the ALS-specific findings themselves.
The researchers noted that DiSPAH’s framework could in principle be applied to other chronic and neurodegenerative diseases where patients show similarly wide variation in both rate and pattern of decline.
What Comes Next
The Nagoya University team acknowledges that DiSPAH requires further validation across a broader range of ALS subtypes, including bulbar-onset ALS — where symptoms begin in speech and swallowing muscles — before it can be widely adopted in clinical settings. The current study focused exclusively on limb-onset cases.
Future research will also explore whether DiSPAH’s predictions remain accurate across ethnically diverse patient populations, as the training datasets used in this study were primarily drawn from existing clinical registries.
Despite these limitations, the research represents a meaningful step toward a future where an ALS diagnosis comes with a personalized roadmap — rather than a single, uncertain prognosis.
The study, titled “DiSPAH: Disentangling Speed and Pathway Heterogeneity in ALS Progression,” was published in npj Digital Medicine. The research was conducted by scientists at Nagoya University’s Graduate School of Medicine.
