Evaluating machine learning classifiers for glaucoma referral decision support in primary care settings.

Several artificial intelligence algorithms have been proposed to help diagnose glaucoma by analyzing the functional and/or structural changes in the eye. These algorithms require carefully curated datasets with access to ocular images. In the current study, we have modeled and evaluated classifiers to predict self-reported glaucoma using a single, easily obtained ocular feature (intraocular pressure (IOP)) and non-ocular features (age, gender, race, body mass index, systolic and diastolic blood pressure, and comorbidities). The classifiers were trained on publicly available data of 3015 subjects without a glaucoma diagnosis at the time of enrollment. 337 subjects subsequently self-reported a glaucoma diagnosis in a span of 1-12 years after enrollment. The classifiers were evaluated on the ability to identify these subjects by only using their features recorded at the time of enrollment. Support vector machine, logistic regression, and adaptive boosting performed similarly on the dataset with F1 scores of 0.31, 0.30, and 0.28, respectively. Logistic regression had the highest sensitivity at 60% with a specificity of 69%. Predictive classifiers using primarily non-ocular features have the potential to be used for identifying suspected glaucoma in non-eye care settings, including primary care. Further research into finding additional features that improve the performance of predictive classifiers is warranted.

Identifying the Critical Factors Governing Translaminar Pressure Differential Through a Compartmental Model.

Purpose: The effective management of glaucoma is hindered by an incomplete understanding of its pathologic mechanism. While important, intraocular pressure (IOP) alone is inadequate in explaining glaucoma. Non-IOP-mediated risk factors such as cerebrospinal fluid (CSF) pressure have been reported to contribute to glaucomatous optic neuropathy. Due to the difficulty associated with experimental measurement of the salient variables, such as the retrobulbar CSF pressure, porosity of the subarachnoid space (SAS), and especially those concerned with the perioptic SAS, there remains a limited understanding of the CSF behavior contributing to the translaminar pressure gradient (TLPG), hypothesized to be a critical factor in the development of glaucoma. Method: An integrated compartmental model describing the intracranial and orbital CSF dynamics, coupled with intraocular dynamics, is developed based on first principles of fluid mechanics. A sensitivity analysis is performed to identify anatomic characteristics that significantly affect the retrobulbar subarachnoid space (RSAS) pressure and, consequently, the TLPG. Results: Of the 28 parameters considered, the RSAS pressure is most sensitive to CSF flow resistance in the optic nerve SAS and the potential lymphatic outflow from the optic nerve SAS into the orbital space. A parametric study demonstrates that a combination of resistance in the range of 1.600 × 1012 - 1.930 × 1012 Pa s/m3 (200.0 - 241.3 mm Hg min/mL) with 5% to 10% lymphatic CSF outflow yields RSAS pressures that are consistent with the limited number of studies in the literature. Conclusions: The results suggest that a small percentage of lymphatic CSF outflow through the optic nerve SAS is likely. In addition, flow resistance in the orbital CSF space, hypothesized to be a function of patient-specific optic nerve SAS architecture and optic canal geometry, is a critical parameter in regulating the RSAS pressure and TLPG.