Artificial intelligence and machine learning for Alzheimer's disease: let's not forget about the retina.

As the world population ages, it is estimated that the population worldwide above the age of 65 years old will increase from 420 million in 2000 to almost 1 billion by 2030.1 Dementia, with Alzheimer’s disease (AD) as the leading cause, is expected to rise in tandem. AD accounts for 60%–80% of all dementia cases,2 with an estimated 5–7 million new cases diagnosed each year.3 Despite intensive research, the diagnosis of AD is currently made through a combination of clinical assessment, neuroimaging and detection of biomarkers from positron emission tomography or cerebrospinal fluid examination,4 with patients facing issues including high costs, invasiveness of the procedures.5 Hence, alternative identification of AD without the use of costly or invasive tests remains a challenge that is difficult to surmount. To date, the healthcare has experienced a significant shift towards early accurate detection as well as early prevention. This importance is highlighted by the screening and surveillance of prevalent diseases such as diabetic retinopathy,6 breast cancer7 and dementia.8 While some of these programmes have been very successful in significantly reducing morbidity and mortality, significant amount of manpower, time and training is required for their successful execution.9 10 This has lent greater weight to the adoption of healthcare technology in order to optimise the accuracy and efficiency of such programmes. Artificial intelligence (AI), through the combination of digitised big data and computational power, has emerged at the forefront of healthcare.11 It appears to be well-suited to address the needs of the healthcare system: fast and accurate predictive, diagnostic and possibly therapeutic algorithms. Machine learning …