Genomics and AI set to transform rare disease diagnosis and therapeutics

Dr Rahila Sardar, Co-founder and CEO, Vgenomics India talks about the role of AI and genomics in diagnosing the rare diseases

More than 350 million people worldwide suffer from over 7,000 rare diseases (RDs). If all rare diseases patients lived in one country, it would be the world’s 3rd most populous country. Approximately 80 percent of rare diseases are genetic, emphasising the role of genetic testing for accurate diagnosis.

Rare diseases pose major hurdles such as delayed diagnosis, limited therapeutic responses, and challenges in monitoring. These difficulties arise due to symptom complexity, insufficient medical expertise, and inadequate diagnostic tools.

In India, an estimated 70 million people live with rare diseases, a figure that cannot be ignored. To address this significant health concern, the Government introduced the National Policy for Rare Diseases (NPRD) in 2021. The policy aims to improve early diagnosis, encourage research, and provide financial assistance of up to Rs 50 lakh through the Rashtriya Arogya Nidhi for eligible patients.

Artificial intelligence (AI) has garnered significant attention for its potential in healthcare, particularly in rare disease detection. By analysing massive datasets, identifying patterns, and integrating clinical information, AI enhances diagnostic accuracy, treatment options, and patient outcomes.

Accelerating diagnosis with AI enhanced genomics

AI systems designed for phenotypic and genomic analysis have proven beneficial in diagnosing rare diseases. The introduction of next-generation sequencing (NGS) approximately a decade ago revolutionised diagnostic processes. Short-read sequencing (SRS) technologies, including exome and genome sequencing, are now standard diagnostic procedures for rare disorders. Next-generation sequencing (NGS) is effective for the diagnosis of rare diseases with a diagnostic yield in the range of 25-50 per cent.

Helping with image identification, genetic analysis, and clinical decision makes AI a crucial tool for the diagnosis of RDs. Machine learning (ML), a form of AI, detects lung involvement more accurately than high-resolution computed tomography (HRCT) and pulmonary function tests (PFTs), currently the standard diagnostic methods. Deep learning (DL), a subset of ML techniques, simplifies complex mappings into clusters of smaller mappings, facilitating efficient analysis.

Artificial intelligence systems utilising genetic data

AI tools efficiently identify gene pathogenicity, discover molecular markers, and develop predictive models based on gene expression data. Tools such as Google’s deepmind, REVEL, CADD are widely used for variant interpretation and pathogenicity predictions.

The Disease-Associated Genome approach, combining phenotypic information with genetic data, has shown significant success in diagnosing Mendelian disorders. This technology examines the genomic sequences, detects variants, and prioritizes them according to pathogenicity and phenotype similarity. Such approaches exemplify AI’s capability in rapidly analysing extensive datasets and identifying relevant variants.

AI technologies also employ statistical methods like Bayesian networks to detect patterns, differentiating and classifying rare versus non-rare disease types, further supporting early identification.

Reducing time to rare disease diagnosis

Patients with rare diseases typically undergo multiple consultations with various specialists over an average span of five years before getting a diagnosis. Emerging studies indicate that AI-assisted diagnostic platforms can significantly reduce diagnostic timelines, in some early pilot cases by up to 10 to 15 times.

Advanced platforms integrating AI algorithms, natural language processing (NLP), and medical data analytics identify patterns associated with rare diseases. These platforms provide risk scores, highlight potential diagnoses, and recommend subsequent actions, such as further testing or specialist referral. Tools like Face2Gene (FDNA) and GestaltMatcher utilise convolutional neural networks to match facial phenotypes to known genetic disorders.

Beyond diagnosis: AI in therapeutic development

Genomics and AI are reshaping the future of rare disease therapeutics. Identifying precise genetic mutations responsible for conditions enables personalised treatments. AI accelerates drug discovery, identifies therapeutic targets, and facilitates drug repurposing. Patients no longer need to wait decades for suitable therapy; precision medicine powered by technology offers new hope.

Accelerating drug discovery and development: AI significantly enhances the initial stages of rare diseases medicine development by swiftly processing data, analysing genetic mutations, and suggesting personalised therapies. For instance, the drug Nitisinone, initially approved for tyrosinemia, was successfully repurposed for Alkaptonuria using AI-supported modelling.

Transforming patient identification – AI showcases transformative potential in patient identification, overcoming persistent limitations of conventional methods.

Improving market access and treatment adoption – AI evaluates the real-world effectiveness of rare disease medications by analysing treatment outcomes, patient quality-of-life improvements, and healthcare utilization changes. It also creates precise budget impact models by integrating demographic data, disease prevalence, treatment trends, and cost analyses.

While AI tools continue to evolve rapidly, their effectiveness relies on robust data, rigorous validation, and clinician oversight, areas that are undergoing continuous improvement.