Autoimmune diseases comprise a spectrum of disorders characterized by the dysregulation of immune tolerance, resulting in tissue or organ damage and inflammation. Their prevalence has been on the rise, significantly impacting patients' quality of life and escalating healthcare costs. Consequently, the prediction of autoimmune diseases has recently garnered substantial interest among researchers. Despite their wide heterogeneity, many autoimmune diseases exhibit a consistent pattern of paraclinical findings that hold predictive value. From serum biomarkers to various machine learning approaches, the array of available methods has been continuously expanding. The emergence of artificial intelligence (AI) presents an exciting new range of possibilities, with notable advancements already underway. The ultimate objective should revolve around disease prevention across all levels. This review provides a comprehensive summary of the most recent data pertaining to the prediction of diverse autoimmune diseases and encompasses both traditional biomarkers and the latest innovations in AI.
Autoimmune diseases (AIDs) represent intricate and heterogeneous disorders arising from a mixture of genetic predisposition, environmental factors, and immune system dysregulation [1]. The consequential damage to tissue and organs, coupled with inflammation [2,3], imposes a significant burden on health care costs [4,5]. Alarmingly, the incidence of AIDs is steadily escalating, influenced by various factors such as dietary habits, air pollution, and infections [6]. What were once considered rare conditions have now exhibited a prevalence of - 5 to 10% across different countries [7,8]. The global scientific community has thoroughly pursued an understanding of AIDs, encompassing their pathogenesis and distinctive behaviors across diverse populations [[9], [10], [11], [12]]. These endeavors have culminated in a valuable reservoir of knowledge, serving as the base for breakthroughs in research pertaining to diagnosis, treatment, and prediction.
Over the past decade, the dominion of artificial intelligence (AI) has experienced remarkable advancements. Subdomains such as machine learning (ML) and deep learning have demonstrated their capacity to predict diseases [[13], [14], [15]]. For instance, a research cohort focused on breast cancer engineered a deep learning model reliant on mammography, outperforming established clinical risk models [13]. Such instances have motivated researchers to seek analogous tools for predicting disorders like AIDs. Yet, AIDs present unique complexities that challenge predictive efforts. Firstly, classifying certain diseases as autoimmune poses difficulties [6,[16], [17], [18]]. Secondly, the role of certain autoantibodies and their pathogenicity remains unclear [19,20]. Thirdly, some antibodies are detectable in both healthy individuals and patients afflicted by non-autoimmune conditions [21].
Despite these obstacles, notable progress has been achieved in predicting AIDs. Biomarkers, molecules whose presence correlate with disease intensity or other pathophysiological states [22], persist as fundamental pillars. However, novel approaches and methodologies have emerged, as we will explore into below. This article reviews the latest insights encompassing antibodies, genetics, and epigenetics findings concerning the prediction of pivotal AIDs. Additionally, we will review potential strategies, including those empowered by AI, designed to anticipate the onset of these disorders.
AI is a broad term that encompasses different disciplines. Fundamental terminology related to AI and examples of ML methods are presented in Table 1 [[23], [24], [25], [26]] and Table 2 [[23], [24], [25], [26]], respectively, and broadly depicted in Fig. 1. In addition, Fig. 2, Fig. 3 represent the process of AI algorithms and ML models to predict outcomes, and Fig. 4 depicts artificial neural networks.
Check access to the full text by signing in through your organization.
SLE is a highly complex disease. Multiple pathogenic changes may occur before the disease becomes clinically evident. For example, dysfunctional DNAses make nuclear antigens, such as double-stranded DNA (dsDNA), accessible to the immune system. These nuclear antigens are then recognized by specific endosomal sensors -Toll-like receptors (TLRs) in plasmacytoid dendritic cells, leading to activation of the Type I Interferon pathway. This induces the secretion of tumor necrosis factor-alpha
Genetics plays a significant role in autoimmune diseases, and understanding the genetic factors involved can contribute to both the diagnosis and prediction of these conditions.
Furthermore, the assessment of specific mRNA expressions emerges as a valuable avenue for predicting and diagnosing autoimmune disorders, particularly in specific clinical settings.
AI tools are now being employed for SLE prediction. For instance, a research team utilized serial serum specimens from 55 individuals collected before SLE classification (average timespan: 4.3 years), and age-matched unaffected healthy controls [137]. They evaluated levels of serum interferon alpha activity, IFN-associated mediators and autoantibodies. Temporal relationships were assessed using random forest models, growth curve modeling, analysis of covariance, and path analysis. Cases were
Our current standards of care may be significantly impacted by AI in a number of ways. Some examples of positive potential include:
Transitioning from prediction to prevention in the context of autoimmune diseases involves considering proactive strategies based on predictive insights. As we describe forecasting models for prediction of several autoimmune conditions, it becomes imperative to shift our focus towards preventive measures. Here, we describe some innovative strategies aimed at averting the onset or progression of autoimmune diseases.
Significant advancements have been achieved in the field of predicting autoimmune diseases. The improved understanding of preclinical stages of these conditions has provided a unique opportunity for early diagnosis and potential disease prevention. The recent FDA approval of a drug that delays the clinical onset of an autoimmune disease represents a groundbreaking development and sets a new standard for therapeutic approaches in this area. The utilization of ML tools is poised to further
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
None.
