Temperament as an Early Risk Marker for Autism Spectrum Disorders? A Longitudinal Study of High-Risk and Low-Risk Infants.

Researchers tracked 200 babies from 6 to 24 months. Half had an older sibling with autism (high-risk); half did not (low-risk).

Parents filled out a 95-item temperament form every six months. The team later checked which children met ASD criteria at 24 months.

High-risk babies scored higher on 'distress to limits' and lower on 'soothability' at the group level. Yet those small differences could not tell you which baby would later be diagnosed.

Only one in five babies who later had ASD had been flagged by extreme temperament scores.

Gur et al. (2024) pushed prediction further. They fed standard wellness-check data into a machine-learning model and correctly spotted about two-thirds of later ASD cases. Their result extends the current work by showing better individual-level accuracy is possible when you add growth and milestone data.

Gunderson et al. (2021) used the same high-risk/low-risk design but looked at sensory responses instead of temperament. They found clear group differences by 12 months, matching the timeline seen here. Together the studies say: group trends show up early, but single-domain measures still fall short for clinical decisions.

Kremkow et al. (2022) systematic review agrees: parent questionnaires give useful group signals, yet digital tools that watch real-time social behavior are edging ahead in proof-of-concept studies.

Temperament forms take five minutes and cost nothing, so they remain a handy first pass for large programs. Just remember they are a weak flashlight, not a spotlight. Use them to decide which families need closer developmental tracking, not to rule autism in or out. Pair the results with milestone charts, sensory checklists, or—where available—automated video or tablet screens for a brighter picture.