use cases

Few Shot Disease Classification

Classify diseases, including rare conditions with few samples, ensuring accurate diagnoses and healthcare.


Accurate disease classification is essential for effective healthcare management, research, and resource allocation. However, the process becomes increasingly challenging and expensive when there is a lack of data, particularly for underrepresented diseases and populations. Inadequate data hinders the development of robust classification models, leading to misdiagnoses, delayed treatments, and insufficient understanding of disease patterns and prevalence. This poses significant challenges for healthcare providers, researchers, and policymakers, limiting their ability to address the specific needs of underrepresented diseases and populations.

Why it matters

  • Insufficient Data Availability: Many diseases, particularly rare or underrepresented conditions, lack sufficient data due to their low prevalence or limited research focus. This scarcity of data makes it difficult to establish accurate and comprehensive disease classification systems. Without a substantial dataset, it becomes challenging to identify distinctive patterns, subtypes, or factors that differentiate one disease from another. The lack of data poses a significant hurdle in developing effective classification models, hindering accurate diagnosis and appropriate treatment selection.
  • Limited Representation of Underrepresented Diseases and Populations: Certain diseases disproportionately affect specific populations, such as ethnic minorities, marginalized communities, or individuals from low-income regions. However, data collection efforts often overlook these populations, resulting in a lack of representation in disease classification systems. Limited representation leads to biased and incomplete models that may not accurately reflect the disease profiles or outcomes within these underrepresented groups. The lack of diverse data restricts the understanding of disease progression, response to treatment, and the development of tailored interventions for these populations.
  • Cost and Resource Intensiveness: Collecting and curating data for disease classification can be a resource-intensive and expensive process. Gathering data from diverse sources, such as electronic health records, clinical trials, or population surveys, requires significant investment in infrastructure, personnel, and technology. For underrepresented diseases and populations, the challenges are amplified due to the scarcity of data sources and the need for targeted data collection efforts. The cost and resource requirements associated with data collection, standardization, and analysis present barriers to developing comprehensive and representative disease classification systems.


To address the challenges of limited data and underrepresentation in disease classification, a solution can be developed using LLM (Language Model) and a custom-made AI system. By leveraging the vast knowledge and language processing capabilities of the LLM, coupled with a tailored AI system, it becomes possible to classify diseases accurately and account for underrepresented diseases and populations. The LLM can be trained on a diverse range of medical literature, clinical guidelines, and patient records to acquire a comprehensive understanding of various diseases. The custom AI system can integrate additional data sources and address specific disease profiles that lack sufficient representation. By harnessing these technologies, healthcare professionals can benefit from improved disease classification accuracy, identification of subtypes, and tailored treatment recommendations. For example, the solution can accurately classify rare diseases by leveraging the collective knowledge of the LLM and complementing it with specialized data collection efforts. By incorporating underrepresented diseases and populations in the training and validation process, the custom AI system can ensure more equitable representation, leading to improved healthcare outcomes for all individuals, regardless of the rarity or underrepresentation of their condition.


  • Historical disease symptoms
  • A few patient examples with correct labels


  1. 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.

Book a Free Consultation

Trusted by the world's top healthcare institutions