Shining a Light on Bone Age: Two Tech Methods Face Off

Figuring out someone’s exact age from their bones is tricky business, especially when the person is no longer growing. For over 40 years, forensic experts have relied on the Suchey-Brooks method—a way to judge adult age by inspecting the pubic symphysis, the small joint where the two halves of the pelvis meet. But now, artificial intelligence is stepping into the ring. This study took 1, 359 pelvic CT scans from a Chinese group and tested two approaches to age prediction. One used the trusty Suchey-Brooks stages, helped along by a math trick called cubic regression. The other let a deep learning model do the heavy lifting through raw number crunching.

Surprisingly, neither method was clearly better. The Suchey-Brooks combo scored errors around six years in both men and women. The deep learning model came in slightly worse, with errors closer to seven years. Both systems struggled most at the extremes—guessing too high for younger adults and too low for older ones. When researchers peeked inside the AI’s decision process, they saw it wasn’t just guessing wildly. It zeroed in on key surface details of the pubic symphysis, though the exact spots changed depending on age and gender. So what does this mean? Neither approach is perfect. Both show age-linked blind spots that affect accuracy. Yet the AI option offers one big plus: speed. Once trained, it can handle hundreds of scans without human fatigue or bias. That makes it a tempting first option for labs racing to turn data into conclusions. Still, no one should trade in the old method just yet—robots trip over the same problems humans do.

Actions