A recent analysis by RARE-X, spanning various knowledge databases like OMIM and Orphanet, reveals a staggering 10,867 rare diseases. Astonishingly, out of this extensive list, only about 500 rare diseases have any FDA approved treatment options against them.[1]

While individually uncommon, approximately 33 million people are grappling with rare diseases in the United States, and globally, nearly 400 million individuals bear the burden of a rare condition. This eclipses the combined patient population of cancer and Alzheimer’s. [2] Remarkably, 20-24% of all cancers fall into the category of rare cancer. [3] A cancer might be deemed rare due to its origin in a distinct subtype of cell, such as T-cell lymphoma, or from an unusual body part like ocular melanoma. Even cancers occurring in a different gender, like breast cancer in men, are classified as rare. All cancers affecting children and teenagers are considered rare. [4]

Rare diseases pose significant challenges for diagnosis and determining the optimal course of treatment. Additionally, due to the limited number of affected individuals, research efforts are hampered, making it difficult to comprehend their etiology and advance drug development. A major hurdle in initiating clinical trials for rare diseases lies in the scarcity of available patient data and challenges in patient enrollment.

Artificial intelligence’s (AI’s) Transformative Role in Rare Disease Research
AI has emerged as a powerful force in accelerating diagnosis, drug development, and treatment optimization for rare diseases.

AI-driven Diagnosis: AI-powered technologies are already enhancing diagnostic speed and accuracy. By automating processes like deep phenotyping, which matches patient phenotype with known Human Phenotype Ontology (HPO), AI complements the broad genetic data generated from whole genome sequencing (WGS). This approach assists previously undiagnosed patients with characteristic phenotypes suggestive of rare diseases, opening doors to medical interventions for those with treatable disorders.

Advertisement #3

Innovative tools like SISH (self-supervised image search for histology) function as a search engine for pathology images, facilitating the diagnosis of rare diseases and unusual conditions that might not have ample cases available. [5]

AI-based Clinical Trial Matching and Patient Recruitment: AI tools are instrumental in projects like NCI-MATCH (National Cancer Institute- Molecular Analysis for Therapy Choice), which automates the matching of tumor molecular signatures to targeted therapies. Physicians receive notifications when a patient’s genomic testing includes an alteration that may render them eligible for participation in the study.[6] Furthermore, AI-based patient recruitment approaches, like patient-centric trial recommendations, enable physicians to input patient data, generating probabilistic matches to available trials. This empowers patients to choose clinical trials that align with their goals and preferences, even if those trials are not enrolling at a local center. [6]

AI-driven platforms are helping scientists accelerate research by efficiently mine existing medical literature and databases, aggregating knowledge about rare diseases. This comprehensive repository speeds up identifying potential treatments and understanding the underlying mechanisms of these conditions. Additionally, AI algorithms facilitate the identification of potential drug candidates through virtual high throughput screening, substantially reducing the time and resources required for drug discovery.

The Challenge of Diversity and Bias
While AI holds immense promise, it is imperative to address the challenge of diversity and potential algorithmic bias. Biomedical datasets are often skewed and under-representative of certain demographic groups. For instance, the representation of Asians, encompassing individuals from the Far East, Southeast Asia, and the Indian subcontinent, in clinical trials conducted from 2000 to 2020, remained consistently low at approximately 1-2%. [7] Notably, the Indian subcontinent alone comprises a quarter of the global population, with Asians representing 60% of the world population, underscoring the potential impact of increased representation in clinical research.

If training data is not representative of population variability, AI can inadvertently reinforce bias, potentially leading to misdiagnoses in historically underrepresented patient groups. This reinforces existing inequalities and must be actively addressed. [8]

Fostering Inclusivity in AI for Healthcare
To ensure new technologies are inclusive and accurate, they must represent the diverse needs of populations. Tackling algorithmic and human bias, bridging information gaps, and establishing data standards and interoperable frameworks are paramount. Implementing the principles of open science into AI design and evaluation tools can facilitate collaboration between the AI and medical fields, amplifying diverse voices in AI deployment for medicine [8][9]

Community-Driven Research and Global Collaboration
Engaging with rare disease communities is critical. By listening to their experiences, understanding their needs, and involving them in the research process, valuable insights can be gleaned. The Indo US Organization for Rare Diseases (IndoUSrare) launched a unique Corporate Alliance program that exemplifies this approach by uniting various life science companies in a pre-competitive spirit to identify and break critical barriers that slow down the development of orphan products. Their mission is to collaborate, educate, empower, and advocate for cross-border and international data sharing and inclusive clinical research on rare diseases globally. Through a virtual patient concierge helping with care navigation and cross-border connections, they are accelerating diagnostics and therapies, bridging critical gaps in data sharing, and fostering affordable clinical research and development.

To tackle the challenges of inclusivity in AI for healthcare, join the conversation at the “Digitization of Rare Diseases” session at the upcoming Indo US Bridging RARE Summit on October 29-30, 2023. Together, we can unlock the full potential of AI in rare disease research.

About Indo US organization for rare diseases:
IndoUSrare is a humanitarian nonprofit 501(c)(3) tax-exempt public charity organization based in the United States. Founder and Executive Chairman Dr. Harsha Rajasimha, who lost a child to a rare disease in 2012, has been a rare disease advocate for more than 10 years. To address the unmet needs of diverse patients with rare diseases globally, the leadership team comprised of experienced professionals from research, advocacy, regulatory, and drug development seeks to build cross-border collaborations connecting stakeholders of rare diseases in low- and middle-income regions such as India, with their counterparts and clinical researchers in the United States to improve the diversity of clinical trial participants, accelerate research and development, and improve equitable access to life-saving therapies to diverse populations of rare disease patients.

[1] The Power of Being Counted. RARE-X report June 2022.[Report]
[2] Navarrete-Opazo AA, Singh M, Tisdale A, Cutillo CM, Garrison SR. Can you hear us now? The impact of health-care utilization by rare disease patients in the United States. Genet Med. 2021 Nov;23(11):2194-2201. doi: 10.1038/s41436-021-01241-7. Epub 2021 Jun 28. PMID: 34183788; PMCID: PMC8553605.
[3] Botta L, Gatta G, Trama A, Bernasconi A, Sharon E, Capocaccia R, Mariotto AB; RARECAREnet working group. Incidence and survival of rare cancers in the US and Europe. Cancer Med. 2020 Aug;9(15):5632-5642. doi: 10.1002/cam4.3137. Epub 2020 May 21. PMID: 32436657; PMCID: PMC7402819.
[4] Rare and less common cancers. Cancer Council Victoria. [Online]
[5] Chen C, Lu MY, Williamson DFK, Chen TY, Schaumberg AJ, Mahmood F. Fast and scalable search of whole-slide images via self-supervised deep learning. Nat Biomed Eng. 2022 Dec;6(12):1420-1434. doi: 10.1038/s41551-022-00929-8. Epub 2022 Oct 10. PMID: 36217022; PMCID: PMC9792371.
[6] Kaskovich S, Wyatt KD, Oliwa T, Graglia L, Furner B, Lee J, Mayampurath A, Volchenboum SL. Automated Matching of Patients to Clinical Trials: A Patient-Centric Natural Language Processing Approach for Pediatric Leukemia. JCO Clin Cancer Inform. 2023 Jul;7:e2300009. doi: 10.1200/CCI.23.00009. PMID: 37428994.
[7] Turner BE, Steinberg JR, Weeks BT, Rodriguez F, Cullen MR. Race/ethnicity reporting and representation in US clinical trials: a cohort study. Lancet Reg Health Am. 2022 Jul;11:100252. doi: 10.1016/j.lana.2022.100252. Epub 2022 Apr 10. PMID: 35875251; PMCID: PMC9302767.
[8] Norori N, Hu Q, Aellen FM, Faraci FD, Tzovara A. Addressing bias in big data and AI for health care: A call for open science. Patterns (N Y). 2021 Oct 8;2(10):100347. doi: 10.1016/j.patter.2021.100347. PMID: 34693373; PMCID: PMC8515002.
[9] Decherchi S, Pedrini E, Mordenti M, Cavalli A, Sangiorgi L. Opportunities and Challenges for Machine Learning in Rare Diseases. Front Med (Lausanne). 2021 Oct 5;8:747612. doi: 10.3389/fmed.2021.747612. PMID: 34676229; PMCID: PMC8523988.

Featured image by Hitesh Choudhary on Unsplash. Used with permission.

Advertisement #5