Using the PDD Behavior Inventory as a Level 2 Screener: A Classification and Regression Trees Analysis.

The team fed 264 parent PDDBI forms into a free CART program. CART means the computer keeps splitting the data to find the best yes-no questions.

Kids were with ASD, ADHD, language delay, or typical development. The goal was to see if the math tree could sort ASD from non-ASD better than the usual cut-off score.

The simple tree hit 81 % accuracy. It used only six PDDBI sub-scales, so you can skip the rest.

The same data also popped out three clear ASD sub-types: language-heavy, social-plus-attention, and sensory-motor.

Rojahn et al. (2012) trimmed the 49-item BPI-01 down to 30 items and kept 0.92-0.99 sensitivity. Tonnsen et al. (2016) does the same trick: fewer items, same punch.

Sappok et al. (2013) warned that ADOS over-calls ASD in adults with ID. Tonnsen et al. (2016) shows CART on PDDBI keeps specificity high, so you dodge that trap.

Rojahn et al. (2012) found SRS over-finds ASD (high sensitivity, low specificity). Tonnsen et al. (2016) gives you a tool that does the opposite: good at ruling in, not just ruling out.

You can paste PDDBI scores into the free CART file right now. In five minutes you get an 81 % accurate ASD flag plus a behavior subtype. That means faster triage and clearer treatment targets without buying new kits.