Table 4. Comparison of More Than 90% Algorithm-Assigned Probability of Exfoliation Syndrome With Clinician-Indicated Absence of Exfoliation Syndrome.

Patient With Discordant Scores	Basis for Algorithm-Assigned Probability Score	Independent Clinician Assessmenta
Case 1	Evidence of XFS in OH and lens examination; however, in the A/P section the clinician specified glaucoma type as “pigment dispersion.”	Patient likely had XFS
Case 2	Evidence of XFS in the lens examination and NLP but not in the problem list or billing codes.	Patient likely had XFS
Case 3	Patient had 21 total visits. XFS was documented in several of the earlier visits prior to the patient undergoing cataract surgery. After cataract surgery, the patient was under the care of a different clinician who instead classified the patient as open-angle glaucoma.	Patient likely had XFS
Case 4	The clinician billed the patient as XFS, but the demographic profile of the patient and clinical exam findings were all consistent with pigment dispersion instead.	Patient likely did not have XFS
Case 5	A glaucoma specialist noted increased pigmentation and speculated that he patient may have XFS but stated in the progress note she was not completely sure of this.	It is possible the patient had XFS
Case 6	There was evidence of XFS in the clinical examination and NLP but not in the problem list or billing codes. The ophthalmology records were mixed in with dozens of non-ophthalmology encounters, which may have made it difficult for the grader to locate evidence of XFS.	Patient likely had XFS

Abbreviations: A/P, assessment and plan; NLP, natural language processing; OH, ocular history; XFS, exfoliation syndrome.

^aAssessment performed by Joshua D. Stein, MD, MS.