Good eyesight belongs to the most-valued attributes of health, and diseases of the eye are a significant health care burden. Case numbers are expected to further increase in the next decades due to an aging society. The development of drugs in ophthalmology, however, is difficult due to limited accessibility of the eye, in terms of drug administration and in terms of sampling of tissues for drug pharmacokinetics (PK) and pharmacodynamics (PD). Ocular quantitative systems pharmacology (QSP) models provide the opportunity to describe the distribution of drugs in the eye as well as the resulting drug-response in specific segments of the eye. In particular, ocular physiologically-based pharmacokinetic (PBPK) models are necessary to describe drug concentration levels in different regions of the eye. Further, ocular effect models employing molecular data from specific cellular systems are needed to develop dose-response correlations. We here describe the current status of PK/PBPK as well as PD models for the eye and describe cellular systems, data repositories as well as animal models in ophthalmology. The application of the various concepts is highlighted for the development of new treatments for post-operative fibrosis after glaucoma surgery.

Accurate transfer learning of clinical outcomes, e.g., of the effects and side effects of drugs or other interventions, from one cellular context to another (in-vitro versus ex-vivo versus in-vivo, or across tissues), between cell-types, developmental stages, omics modalities or species, is considered tremendously useful. Ultimately, it may avoid most drug development failing in translation, despite large investments in the preclinical stages, which includes animal experiments requiring careful justification. Thus, when transferring a prediction task from a source (model) domain to a target domain, what counts is the high quality of the predictions in the target domain, requiring molecular states or processes common to both source and target that can be learned by the predictor, reflected by latent variables. These latent variables may form a compendium of knowledge that is learned in the source, to enable predictions in the target; usually, there are few, if any, labeled target training samples to learn from. Transductive learning then refers to the learning of the predictor in the source domain, transferring its outcome label calculations to the target domain, considering the same task. Inductive learning considers cases where the target predictor is performing a different yet related task as compared to the source predictor, making some labeled target data necessary. Often, there is also a need to first map the variables in the input/feature spaces (e.g. of gene names to orthologs) and/or the variables in the output/outcome spaces (e.g. by matching of labels). Transfer across omics modalities also requires that the molecular information flow connecting these modalities is sufficiently conserved. Only one of the methods for transfer learning we reviewed offers an assessment of input data, suggesting that transfer learning is unreliable in certain cases. Moreover, source domains feature their very own particularities, and transfer learning should consider these, e.g., as differences in pharmacokinetics, drug clearance or the microenvironment. In light of these general considerations, we here discuss and juxtapose various recent transfer learning approaches, specifically designed (or at least adaptable) to predict clinical (human in-vivo) outcomes based on molecular data, towards finding the right tool for a given task, and paving the way for a comprehensive and systematic comparison of the suitability and accuracy of transfer learning of clinical outcomes.

The most important predictors for outcomes after ischemic stroke, that is, for health deterioration and death, are chronological age and stroke severity; gender, genetics and lifestyle / environmental factors also play a role. Of all these, only the latter can be influenced after the event, even though recurrent stroke may be prevented by antiaggregant/anticoagulant therapy, angioplasty of high-grade stenoses, and treatment of cardiovascular risk factors. Moreover, blood cell composition and protein biomarkers such as C-reactive protein or interleukins in serum are frequently considered as biomarkers of outcome. We surveyed protein biomarkers that were reported to be predictive for outcome after ischemic stroke, specifically considering biomarkers that predict long-term outcome (≥3 months) and that are measured over the first days following the event. We classified the protein biomarkers as immune‑inflammatory, coagulation-related, and adhesion-related biomarkers. Some of these biomarkers are closely related to cellular senescence and, in particular, to the inflammatory processes that can be triggered by senescent cells. Moreover, the processes that underlie inflammation, hypercoagulation and cellular senescence connect stroke to cancer, and biomarkers of cancer-associated thromboembolism, as well as of sarcopenia, overlap strongly with the biomarkers discussed here. Finally, we demonstrate that most of the outcome-predicting protein biomarkers form a close-meshed functional interaction network, suggesting that the outcome after stroke is partially determined by an interplay of molecular processes relating to inflammation, coagulation, cell adhesion and cellular senescence.

Despite increasing research efforts, there is a lack of consensus on defining aging or health. To understand the underlying processes, and to foster the development of targeted interventions towards increasing one’s health, there is an urgent need: (1) to find a broadly acceptable and useful definition of health, based on a list of features (which may or may not be molecular); (2) to operationalize features of health so that it can be measured; (3) to identify predictive biomarkers and (molecular) pathways of health, and (4) to suggest interventions, such as nutrition and exercise, targeted at putative causal pathways and processes. Based on a survey of the literature, we propose to define health as a state of an individual characterized by the core features of (a) physiological function, (b) cognitive function and (c) physical function, amended, specifically in case of humans, by (d) lack of disease, and by (e) reproductive function. Often used concepts such as lack of frailty, allostatic load, or self-reported health (in case of human), and indices such as the Healthy Aging Index can be viewed as projections or surrogates of our definition. We further define aging as the set of all processes in an individual that reduce its “wellbeing”, that is, its health or survival or both. We define biomarkers of health by their attribute of predicting future health better than chronological age. We define healthspan pathways as molecular features of health that relate to each other, specifically by belonging to the same molecular pathway. Our conceptual framework may integrate diverse operationalizations of health and guide precision prevention efforts that are a key to reducing the need for medical and nursing care.