Optimization of current treatment regimens of dry eye disease through machine learning

The outer part of the tear film consists of a thin lipid layer. This lipid layer protects the eye from its surroundings, delays evaporation of the underlying watery portion, and thus prevents dehydration of the ocular surface. The tear film consists of several layers containing thousands of proteins, lipids and metabolites. Over the past few years, technological advancements have greatly increased the number of identified components in the tear film. Dry eye disease is a result of several underlying mechanism with the common consequence of a destabilized tear film. This project aims to contribute to the search for a more lasting and individualized treatment regime of dry eye disease through artificial intelligence. To achieve this, the following subprojects are planned: 1. Calculate the prevalence of dry eye disease among common eye diseases such as glaucoma, cataract and age related macula degeneration. 2. Determine the ideal treatment regimen for a given subgroup and stage of dry eye disease through artificial intelligence. 3. Determine whether light treatment through Valeda (used for treating age related macula degeneration) might have a therapeutic effect on dry eye disease. 4. Optimize the treatment parameters used in “Intense Pulsed Light” treatment through artificial intelligence to improve subjective and objective clinical findings. All patients will undergo a thorough dry eye examination combined with biochemical analysis prior to and following treatment with the goal of determining the clinical and biochemical effects of the therapy. As this approach generates enormous amounts of data, artificial intelligence will be used to analyse the datasets as objectively as possible. We will also develop models with the aim to discover new connections between diagnostics and different medical treatments, including light therapy, in the hope to gain better clinical results.

The lipid layer of the human tear film functions as a hydrophobic barrier protecting the eye from external agents and prevents evaporation of the aqueous layer and, hence, corneal desiccation. The tear film consists of thousands of proteins, lipids and metabolites. Over the recent years, advances in analytical equipment has helped identify more than 600 lipid species from 17 lipid classes in human meibum and tear film. Dry eye disease (DED) is a multifactorial disease whose core hallmark is the disruption of tear film homeostasis. The present proposal aims to contribute to the search for more personalised and lasting treatment regimens for DED through machine learning methods. To this end, the following projects are planned: 1. We aim to calculate the prevalence of DED among patients with glaucoma, age-related macular degeneration, cataract and other common eye diseases. 2. Determine the ideal treatment regimen concerning subgroups and various stages of DED through machine learning. 3. Determine whether photobiomodulation through Valeda might have a therapeutic effect on DED. 4. Optimize IPL treatment parameters by means of machine learning to improve subjective and objective clinical findings resulting in a less inflammatory tear film. All patients will undergo a thorough dry eye examination as well as proteomic and lipidomic analysis prior to- and after treatment with the goal of determining the subjective, clinical and biochemical effects of the therapy. The present project will apply machine learning to analyse a vast dataset in a completely unbiased manner. Furthermore, we will utilize machine learning to discover novel associations between treatment response to subjective and objective parameters as a result of various treatment protocols and photobiomodulation. To our knowledge, machine learning, despite its exponentially increasing medical applications, has never been applied in this setting before, which increases the innovation potential of the project.