The comprehensive health checkup system is a unique system that supports excellent health status in our country. By conducting large-scale and regular screening tests on an apparently healthy population, the system detects many diseases while asymptomatic and enables early initiation of interventions for prevention and treatment. However, since most of the target population is healthy, some unique problems arise due to the extremely low prevalence of diseases. Even with highly accurate tests, the number of false positives can be significant. Artificial intelligence can automate tasks currently performed by humans and build systems that detect diseases directly from tests, enabling an expansion of the information that can be handled in health checkups and significantly improving the accuracy of tests. This not only reduces the oversight of diseases but also reduces unnecessary secondary tests, cutting medical costs and minimizing unnecessary bodily invasion and radiation exposure. Moreover, machines can perform the same calculations repeatedly without fatigue, making them ideal for large-scale population targets like health checkups.

 While AI is extremely useful, the concerns in public discourse that it will 'steal doctors' jobs' are unlikely to occur any time soon. It is merely akin to obtaining superior test results, and it cannot substitute for the critical work of physicians, such as interpreting results or making subsequent step decisions based on patient values. Physicians, are required to properly understand the strengths and weaknesses of AI and to use it skillfully.

 This paper provides an overview of AI technology evolution and its application in the medical field. The third AI boom, starting in the early 2000s with the advent of deep learning, notably advanced in image recognition. The emergence of the transformer algorithm in 2017 and subsequent development of large language models like ChatGPT marked the fourth AI boom. In medicine, the adoption of AI for image diagnostic support has grown, with numerous medical devices approved by the FDA and PMDA. Future developments are expected in multimodal Large language Models (LLMs), integrating language, images, and other data.

 Recently, there has been a surge in the development of products and services using Artificial Intelligence (AI), with numerous AI-based research and medical device developments occurring in the healthcare sector. In the field of cardiology, there is progress in developing diagnostic support systems utilizing digital health based on vast amounts of data. Particularly in the area of electrocardiogram (ECG) interpretation, many studies report on AI-created automatic ECG diagnostic tools, trained with massive amounts of ECG image data and enhanced image recognition technology. Research is also being conducted to visualize the decision-making process of AI, which has been considered a "black box," making it comprehensible to the human eye. Cutting-edge research is advancing the practical application of technology to detect findings and signs in ECGs, which are normally imperceptible to human observers, using AI. This includes predicting the onset of conditions like left ventricular dysfunction and transient atrial fibrillation. The use of AI to detect disease signs holds promise for efficiently screening high-risk patients before the onset of illness, which is expected to have a significant impact in the field of preventive medicine.

 The number of patients with chronic kidney disease in Japan is estimated to be around 14 million. As the disease progresses, patients not only require dialysis and kidney transplantation, but it is also related to major causes of death due to decreased immunity and progression of arteriosclerosis. In addition to renal biopsy, imaging tests such as ultrasound, MRI and CT are used for diagnosis, but in routine clinical practice, serum creatinine, proteinuria and blood pressure are the main assessment parameters.

 This existing clinical data is used as a feature for AI diagnostics to predict the onset of acute kidney injury, life prognosis, and risk of progression of kidney disease. The introduction of AI techniques into the quantitative evaluation of kidney biopsies is also being explored, demonstrating high accuracy in tissue classification and lesion identification.

 Advances in imaging diagnostics, represented by MRI, have made it possible to visualize renal pathologies that were previously unattainable. Innovative imaging techniques allow for the non-invasive acquisition of information related to renal ischemia, hypoxia, perfusion, and changes in microstructure and fibrosis. AI technology is heavily relied upon as a means to comprehensively and quantitatively process vast amounts of imaging information.

 In the absence of effective treatments for chronic kidney disease, early diagnosis and appropriate therapeutic intervention are critical. The use of AI technology is expected to contribute to accurate diagnosis and evaluation of treatment effects for kidney disease, and to have a significant impact on the progress of clinical research.

 The Japan Brain Dock Society is a unique brain health checkup system in Japan, unprecedented in the world, for early detection and early treatment of asymptomatic brain diseases such as unruptured cerebral aneurysms and asymptomatic cerebral infarctions. Currently, it also focuses on dementia prevention as a medical society for stroke and dementia prevention. Japan has had imaging data from brain health checkups since the 1920s, and can be said to have a treasure trove of brain imaging evidence from healthy individuals. However, not all of this data has been registered in databases, and we are currently accumulating and studying the data that has been compiled into databases. For example, it was thought that subarachnoid hemorrhage would be reduced if many unruptured cerebral aneurysms were detected and treated during Brain Dock. However, it is now known that the number of unruptured cerebral aneurysms detected during Brain Dock is limited and that it is difficult to reduce subarachnoid hemorrhage by treating these aneurysms, and that improving lifestyle through Brain Dock can contribute to reducing the frequency of subarachnoid hemorrhage.

 Artificial intelligence (AI), a method that derives correct answers from supervised data, is suitable for assisting diagnostic imaging in Brain Dock and predicting future risk of cerebral stroke and dementia, and is already in practical use. Currently, AI for measuring brain atrophy, especially hippocampal atrophy, is in practical use, and we are developing AI to assist in the diagnosis of unruptured brain aneurysms and AI for brain white matter changes. On the other hand, in addition to MRI image data, other data such as medical history, lifestyle, blood sampling data, electrocardiogram, carotid ultrasonography, and cognitive function tests are collected during actual brain checkups. It may be possible to integrate these data and make predictions using AI. This paper describes the application of AI to diagnostic imaging support and the possibility of Brain Dock diagnosis by AI.

 Objective: To investigate the effectiveness of fundus photography in specific health checkups by the proportion of patients whose ophthalmology checkups led to a cataract surgery plan and ophthalmologic findings.

 Methods: Fundus photography was performed on 1,928 residents who underwent specific health checkups between June and December 2022, and 193 were selected for a thorough examination. A total of 102 participants who brought their screening forms were included in the analysis. The history of ophthalmology examinations and whether cataract surgery was planned were retrospectively investigated from the medical records. The participants were divided into a first-visit group (first G) and a second-visit group (second G) based on their history of ophthalmologic examinations and were compared.

 Results: The participants were aged 73.6 ± 7.7 years (mean ± standard deviation); 47 (46.1%) were male, and 55 (53.9%) were female. Of those who underwent a thorough examination, 52 (51.0%) had an initial visit, and 50 (49.0%) had a repeat visit to our department. Cataract was diagnosed in 71 patients (69.6% prevalence). The results of the intergroup comparison showed that 42 (41.2%) were diagnosed with cataracts in first G and 29 (28.4%) in second G, and the number of planned surgeries was 20 (38.5%) and 8 (16.0%) in the first and second G, respectively, both with significant biases (p < 0.05, Pearson's chi-square test).

 Conclusions: The usefulness of fundus imaging in health checkups for residents to detect cataracts was evidently demonstrated.