Journal article
Unsupervised Machine Learning Shows Change in Visual Field Loss in the Idiopathic Intracranial Hypertension Treatment Trial
Ophthalmology (Rochester, Minn.), Vol.129(8), pp.903-911
04/01/2022
DOI: 10.1016/j.ophtha.2022.03.027
PMID: 35378137
Abstract
We previously reported that archetypal analysis (AA), a type of unsupervised machine learning, identified and quantified patterns of visual field (VF) loss in idiopathic intracranial hypertension (IIH), referred to as archetypes (ATs). We assessed whether AT weight changes over time are consistent with changes in conventional global indices. We explored whether visual outcome or treatment effects are associated with select ATs and whether AA reveals residual VF defects in eyes deemed “normal” after treatment.
Analysis of data collected from a randomized controlled trial.
2,862 VFs taken from 165 participants during the Idiopathic Intracranial Hypertension Treatment Trial (IIHTT).
We applied a 14-AT model derived from IIHTT VFs. We examined changes in individual AT weights over time within all study eye VFs and evaluated differences between treatment groups. We created an AT Change score to assess overall VF change from baseline. We tested threshold baseline AT weights for association with VF outcome and treatment effect at six months. We determined the abnormal ATs with meaningful weight at outcome for VFs considered “normal” based on a mean deviation (MD) cutoff ≥-2.00 dB.
Individual AT weighting coefficients, MD.
AT1 (a normal VF pattern) showed the greatest weight change for all study eyes, increasing from 11.9% (interquartile range [IQR]: 0.44–24.1%) at baseline to 31.2% (IQR: 16.0–45.5%) at outcome (p<0.001). AT1 weight change (r=0.795, p<0.001) and a global score of AT change (r=0.988, p<0.001) correlated strongly with MD change. Study eyes with baseline AT2 (a mild diffuse VF loss pattern) weight ≥ 44% (≥1 standard deviation above the mean) showed higher AT2 weights at outcome than those with AT2 < 44% at baseline (p<0.001). Only the latter group showed a significant acetazolamide treatment effect. AA revealed residual VF loss patterns, most frequently representing mild diffuse loss and enlarged blind spot in 64 of 66 study eyes with MD ≥-2.00 dB at outcome.
AA provides a quantitative approach to monitoring VF changes in IIH. Baseline AT features may be associated with treatment response and VF outcome. AA uncovers residual VF defects not otherwise revealed by MD.
Archetypal analysis of visual fields (VFs) from the Idiopathic Intracranial Hypertension Treatment Trial facilitates quantification of VF change over time, identifies baseline VF patterns associated with outcome or treatment effect, and reveals residual VF defects.
Details
- Title: Subtitle
- Unsupervised Machine Learning Shows Change in Visual Field Loss in the Idiopathic Intracranial Hypertension Treatment Trial
- Creators
- Hiten Doshi - Albert Einstein College of MedicineElena Solli - Icahn School of Medicine at Mount SinaiTobias Elze - Harvard UniversityLouis R Pasquale - Icahn School of Medicine at Mount SinaiMichael Wall - University of IowaMark J Kupersmith - Icahn School of Medicine at Mount Sinai
- Resource Type
- Journal article
- Publication Details
- Ophthalmology (Rochester, Minn.), Vol.129(8), pp.903-911
- Publisher
- Elsevier Inc
- DOI
- 10.1016/j.ophtha.2022.03.027
- PMID
- 35378137
- ISSN
- 0161-6420
- eISSN
- 1549-4713
- Grant note
- DOI: 10.13039/100001818, name: Research to Prevent Blindness, award: NEI R01 EY015473
- Language
- English
- Date published
- 04/01/2022
- Academic Unit
- Neurology; Ophthalmology and Visual Sciences
- Record Identifier
- 9984240759902771
Metrics
12 Record Views