ENTERPRISE AI ANALYSIS

In the past decade, AI research has significantly improved the forecasting, identification, diagnosis, categorization, treatment, and survival forecasting of diseases (53, 54), fostering medical innovation and promoting a sustainable approach to precision medicine. Biomarker tests or tools indicate normal biological processes, pathogenic processes, or predictive biomarkers are measured once to forecast future events, while monitoring, response, and safety biomarkers are measured over time (55). AI applications have exponentially grown across various fields, with the earliest recorded automated pattern recognition appearing in a 1960 report in The Lancet (56). Precision medicine approaches are already being implemented in the context of cancer, encompassing both diagnosis and treatment. In the realm of cancer diagnosis, current literature (57-59) showcases numerous studies delving into AI’s potential, comparing its results to manual detection by pathologists. AI exhibits superior accuracy compared to human pathologists in diagnosing certain cancer types (60-62). The precision medicine approach tailor’s cancer treatment plans by considering tumor-associated and inherited genetic variations, environmental exposures, lifestyle, general health, and medical history. AI is revolutionizing drug discovery, target identification and clinical trials. Traditional methods used in drug discovery are often expensive and time-consuming, and they may not consistently offer accurate forecasts of a drug’s efficacy and safety. The use of AI algorithms, particularly ML models, has significantly advanced drug discovery through better predictive analytics and target identification. By analyzing vast amounts of data, these algorithms hasten the early stages of drug development by detecting patterns and predicting potential drug candidates (63). AI models, such as those in ML and deep learning, draw on data from genomics, proteomics, chemical structures, and clinical trials to pinpoint drug candidates, evaluate safety, and estimate effectiveness using historical data. By forecasting interactions, toxicity, and pharmacokinetics of compounds, AI allows researchers to prioritize drug candidates for further development (64). Moreover, AI simplifies the process of identifying targets by examining biological data and understanding disease mechanisms. AI-powered strategies, such as network analysis and building knowledge graphs, bring together diverse data sources to highlight promising therapeutic targets (65). AI also integrates multi-omics data effectively, offering a broad understanding of disease pathways and improving target identification accuracy (66). Furthermore, neural networks and other deep learning models rank drug targets by evaluating complex connections between molecular characteristics and disease pathways, aiding in therapeutic intervention (67). To find actionable therapeutic targets in amyotrophic lateral sclerosis (ALS), Pun et al. (68) integrated various bioinformatic and deep learning models, which were trained on disease-specific multitopic and text data to prioritize drug-gable genes, revealing 18 potential targets for ALS treatment. Additionally, West et al. (69) created a deep learning method with an innovative modular structure to pinpoint human genes connected to multiple age-related diseases by studying patterns derived from gene or protein attributes such as Gene Ontology terms, protein-protein interactions, and biological pathways. Through automating the identification of suitable participants, AI also has the potential to improve patient recruitment and eligibility in clinical trials. AI algorithms process electronic health records, medical literature, and additional healthcare data to pinpoint potential candidates. Automated prescreening tools help to minimize manual tasks and increase efficiency in recognizing eligible participants (70). Predictive analytics improve recruitment by predicting enrollment rates, enabling trial sponsors to allocate resources efficiently using patient demographics and historical data (71). AI is also altering clinical trial approaches with the help of real-world evidence (RWE) and adaptive trial designs. AI algorithms examine data from various real-world sources to offer a more comprehensive insight into patient demographics, disease development, and treatment results (72). AI supports the division of patients into categories and the recognition of subgroups, targeting distinct patient profiles for adaptive trials, and predictive analytics help in anticipating recruitment rates and treatment responses (73). By adapting trial designs in real-time, protocols are optimized, leading to greater trial efficiency. In general, AI’s adaptive trial designs increase flexibility, efficiency, and success.

Based on the summary of the Public Health and Epidemiology Informatics section in the 2017 IMIA Yearbook (79), precision public/global health and digital epidemiology are still used in 2018 (80, 81). It entails providing the appropriate intervention to the suitable population at the optimal time (80). The latter phrase pertains to employing digital data, especially data not deliberately gathered, to answer epidemiological questions (81). The significant potential of Big Data in epidemiology was showcased by Deiner et al.’s (82) innovative study, which demonstrated that monitoring social media for disease symptom queries can lead to early detection of epidemics. Pattern recognition and data analytics were employed to detect, identify, and categorize patterns of disease occurrence associated with conjunctivitis. Conversely, wearable technologies will enable the monitoring and collection of individual medical information and the refinement of the care process. The fusion of AI with virtual reality and augmented reality (83), will enable the creation of both virtual medical services that citizens can access easily and directly, as well as increasingly effective and safe applications for robotic surgery. It is noteworthy that the keyword "ethics" is depicted in Figure 10, indicating a growing focus on AI ethics in medicine. Regulatory laws and guidelines for medical AI are frequently formulated without engaging in dialogue among community members, clinicians, developers, and ethicists. This lack of collaboration may result in regulations that do not align with the experiences of community members as users of medical AI. Ethical concerns highlighted by policymakers and scholars may not match those of patients, providers, and developers, leading to a disconnect that makes ethical decision-making tools ineffective for AI users. The ethical issues identified by policymakers and scholars may not correspond with those of patients, providers, and developers, creating a disconnect that makes ethical decision-making tools ineffective for AI users (84). Analyzing empirical studies on the ethics of medical AI assists educators, researchers, and ethicists in understanding and addressing perceived ethical concerns (85).