A recent study suggests that machine learning analysis consistently detects glaucoma progression earlier with or without confirmation visits.

A team of researchers analyzed visual fields from 2,085 eyes of 1,214 subjects to identify glaucoma progression patterns using a new machine learning technique. They then used the visual fields from 133 eyes of 71 glaucoma patients as a no-change, test-retest dataset. To compare methods, the team evaluated an independent dataset of 270 eyes of 136 glaucoma patients.

The researchers found that it took 5.2 years to detect progression in 25% of the eyes in the longitudinal dataset using global mean deviation, 4.5 years using region-wise and 3.9 years using point-wise. The shortest detection rate was 3.5 years with machine learning analysis, the study notes.

Further supporting its findings, the team discovered that it only takes 5.1 years to detect progression in 25% of eyes after two additional visits using machine learning, while it takes 6.6 years, 5.7 years and 5.6 years with global, region-wise and point-wise, respectively.

Not only does machine learning analysis detect progressing eyes earlier than other methods, but the study concludes that the technique also helps detect eyes that are progressing at a slower rate.