Prediction of Poor Visual Outcomes at Idiopathic Intracranial Hypertension Diagnosis Using a Supervised Machine...

Area of Science:

• Ophthalmology
• Neurology
• Data Science

Background:

• Idiopathic intracranial hypertension (IIH) is a vision-threatening condition primarily affecting women of reproductive age.
• Early diagnosis and intervention are crucial for preventing vision loss, yet validated predictive tools are lacking.
• This study aimed to develop a machine learning algorithm for predicting visual outcomes in IIH patients at diagnosis.

Purpose of the Study:

• To create a machine learning algorithm to predict poor visual outcomes in patients diagnosed with IIH.
• To stratify patients into risk groups based on their likelihood of vision loss.
• To identify key prognostic markers for visual outcomes in IIH.

Main Methods:

• A retrospective cohort study analyzed electronic health records of 391 IIH patients (aged 0-70) diagnosed between June 2012 and September 2023.
• Poor visual outcomes were defined as visual field mean deviation (VFMD) worse than -7 dB or visual acuity of 20/80 or worse.
• Decision tree and logistic regression models were developed and evaluated using accuracy, sensitivity, specificity, and AUC, with k-fold cross-validation for validation.

Main Results:

• Decision tree models demonstrated superior performance, leading to the creation of four prognostic risk groups: critical, high, medium, and low.
• Patients in the critical risk group with VFMD worse than -12.59 dB and non-White ethnicity had a 92.6% risk of poor visual outcome.
• A baseline VFMD worse than -9.1 dB indicated a 69.8% critical risk of poor visual outcome, while VFMD better than -3.39 dB had only a 1.04% risk.

Conclusions:

• The study provides clinicians with prognostic markers to identify IIH patients at critical risk of significant vision loss.
• A VFMD worse than -9.1 dB signifies critical risk for poor visual outcomes.
• Minority patients with a VFMD worse than -9.1 dB face an even higher risk of vision loss.