Unraveling Mechanisms of Disease Pathogenesis with AI

The journey from bench to market fraught with uncertainty, with 80% of the cost of bringing a drug to market spent on molecules that ultimately fail.¹ The lack of disease relevance of a target or mechanism contributes to such failures, leading to safety concerns or drugs with a narrow therapeutic index.² The key to success in drug development lies in understanding genes, proteins and pathways involved in the main pathophysiological processes.³ Such insights are crucial for stratifying patients most appropriate for treatment via a given mechanism, as well as identifying the right molecular targets.⁴ A foundational understanding of disease pathogenesis therefore serve as a roadmap for success.

Navigating the landscape of pathophysiological mechanisms is challenging due to the complexity and heterogeneity of diseases and the influence of genetic, environmental, and lifestyle factors. This is for exacerbated by the rapidly expanding volume of scientific literature, growing at an annual rate of around 9%.⁵ To overcome these challenges, pharma companies are leveraging AI for scientific knowledge extraction. AI models can efficiently sift through large datasets, identify patterns and extraction relevant information, thereby facilitating a more comprehensive understanding of disease pathogenesis.

In this use case, we present an example workflow showing how Causaly’s human-centric AI extracts meaningful insights from the entirety of biomedical literature to help explore mechanisms of disease pathogenesis (Figure 1).

1. Identifying Biomarkers for Fibromyalgia

Using Causaly, 450+ biomarkers of fibromyalgia disease progression were extracted from almost 700 relevant publications. Interluekin-6 (IL6) was identified as the most studied biomarker, with 237 hyperlinked articles supporting how IL6 can regulate the progression of fibromyalgia. While investigating further the relationship of IL6 with this disease, 67 documents provided direct evidence to support the biomarker-disease relationship.

To focus on primary findings, biomarkers were filtered by results, discussion and conclusion article sections, narrowing the search down to 23 publications. From this, we can see exact sentences from texts where IL6 has been implicated in the pathogenesis of fibromyalgia.

2. Uncover Mutually Associated Biochemical Pathways

To better understand biological mediators and signaling cascades affected by IL6 as it relates to fibromyalgia, we used Causaly’s multihop module to uncover almost 340 potential IL6-affected biochemical pathways that are mutually associated to fibromyalgia. This unveiled a majority of relationships connecting IL6 with increased JAK activity, where evidence points to where JAK pathway plays a role in fibromyalgia.

3. Explore Cellular Targets Affecting Disease Pathogenesis

Using this basis for further elucidation of under-studied signaling pathways potentially involved in fibromyalgia pathogenesis, we can further understand the implicated cellular targets by which the JAK pathway mediates disease pathogenesis. To do this, we can perform an iterative multihop search. This revealed ~60 implicated cells linking JAK signaling to fibromyalgia. For example, the inhibition of JAK has shown to attenuate the immunosuppressive capacity of mesenchymal stem cells (MSCs).⁶

4. Generate Hypotheses with Confidence

Building on these insights, we can go back to Causaly’s intelligent search to further explore the relationship between MSCs and fibromyalgia. This unveiled 74 MSC-associated genes and proteins which affect fibromyalgia.

From these investigations, we can formulate ideas about how IL6 affects fibromyalgia development via JAK signaling. Moreover, we can hypothesize downstream effects of genes and proteins that may be implicated in the development of fibromyalgia, which may serve as potential targets or off-targets as a consequence of certain treatments.

AI-supported investigations into disease mechanisms offer a transformative approach to understand complex pathologies, such as fibromyalgia. By leveraging AI to unlock disease pathogenesis understanding, researchers can identify novel biomarkers and therapeutic targets with unprecedented precision. This not only accelerates the journey from lab to market but also enhances the efficacy and safety of new treatments.