Alzheimer’s disease (AD) is the most common neurodegenerative disorder and has complicated pathobiology, leading to irreversible memory loss and severe cognitive dysfunction. For patients with AD, the advent of the disease usually occurs after years of pathological changes. The early diagnosis and monitoring of AD are of great significance as the early-stage intervention and treatment may be the most effective. Biomarkers, such as beta-amyloid and tau levels in cerebrospinal fluid (CSF) and brain, offer one of the most promising paths and are combined with neuroimaging and immunological detection for AD diagnosis. However, high expense and radiation of neuroimaging and low sensitivity of immunosorbent assay limited their applications. Meanwhile, the relevance of Aβ peptides and tau proteins to the development of AD remains highly debatable, meaning that detecting one specific biomarker holds limited prospects in achieving early and accurate detection of AD. Optical sensor arrays based on pattern recognition enable the discrimination of multiple analytes in complicated environments and are thus highly advantageous for the detection of AD with multi-biomarkers. In this review, we survey the recent advances of optical sensor arrays for the diagnosis of AD, as well as the remaining challenges.