Princeton: A large study has identified four distinct patterns within autism, offering new evidence about why the condition can present differently from one person to another.

Researchers from Princeton University, the Flatiron Institute and the Simons Foundation examined data from 5392 autistic children aged four to 18 who participated in the SPARK autism research programme.

Instead of studying one characteristic at a time, the team used a computational model to analyse 239 features together. These included social interaction, communication, repetitive behaviours, mental health conditions and the ages at which children reached developmental milestones.

The model identified four groups.

The Social and Behavioural Challenges group represented about 37 per cent of participants. Children in this group generally reached developmental milestones on time but often experienced communication difficulties, repetitive behaviours, anxiety, depression or attention deficit hyperactivity disorder.

The Mixed Autism Spectrum Disorder with Developmental Delay group accounted for about 19 per cent. These children commonly reached milestones such as walking and speaking later but showed fewer anxiety, mood, or disruptive behaviour conditions.

The Moderate Challenges group included approximately 34 per cent of participants. Children showed core autistic characteristics at a lower intensity, generally reached developmental milestones on time and had fewer accompanying psychiatric conditions.

The Broadly Affected group was the smallest at about 10 per cent. It included children with wider developmental, communication and behavioural challenges as well as a greater frequency of mental health conditions.

Genetic analysis also revealed differences among the groups. The Broadly Affected group had the highest proportion of harmful genetic changes that were not inherited from either parent. The group with developmental delay showed a greater likelihood of carrying rare inherited variants.

The researchers also found that affected genes appeared to become active at different stages of development depending on the group.

Published in Nature Genetics in July 2025, the study does not replace the existing autism diagnosis or prove that every autistic person belongs permanently to one of four categories.

The researchers said the results establish at least four meaningful patterns, but further studies involving more diverse populations are required.

If validated, the framework could eventually help clinicians anticipate support needs, monitor related conditions and develop more personalised care.