← Other projects

RDD [Rare Disease Discovery]: improving the initial diagnosis of rare diseases

Initial diagnosis of rare diseases by family doctors can be difficult due to the overlapping of symptoms among sets of rare diseases and between rare diseases and other more common ones. Computationally assisted diagnosis (CAD), based on the patient's symptoms, represents a potentially rapid, convenient and inexpensive solution to help in this initial diagnosis.

Some computer tools have been designed specifically for the initial CAD of rare diseases from the symptoms of the patient. One of them is RDD [Rare Disease Discovery, http://disease-discovery.udl.cat/], a prototype developed in our group, which has a high ease of use and speed of execution. The diagnostic performance of RDD in large-scale tests indicates that RDD has better diagnostic precision and sensitivity than other CAD tools for rare diseases.

RDD has been given the Research Group Award 2018 by the Lleida section of the Junior Chamber International (JCI), an important organization that is present in more than 100 countries. JCI is a nonprofit organization of young active citizens age 18 to 40 who are engaged and committed to creating impact in their communities. Active citizens are individuals invested in the future of our world. JCI gathers active citizens from all sectors of society. We develop the skills, knowledge and understanding to make informed decisions and take action.

In spite of its promise, there are several important limitations in the applications that are specific for the CAD of rare diseases. We highlight four. 

  1. First, those applications do not provide a differential diagnosis between common and rare diseases; they only provide a differential diagnosis amongst rare diseases.
  2. Second, a large-scale, comprehensive study with real patients is needed to better gauge their accuracy and sensitivity.
  3. Third, the best statistical and mathematical method to provide a reliable initial diagnosis of rare diseases is still to be clarified.
  4. Finally, no CAD method for rare diseases has the ability to predict a rare disease that has not been previously observed.

This project continues the development of the RDD tool, aiming at providing solutions to the four issues described in the previous paragraph.

  1. First, RDD will include a functionality to estimate the probability that the list of symptoms of a patient is associated with either a common disease or a rare disease.
  2. Second, we will validate the various stages of the tool by using appropriately anonymized information from international sources with the symptoms of more numerous groups of patients with rare diseases.
  3. Third, we will further test and develop the computational methods for assisted diagnosis based on the symptoms, with the aim of finding the most reliable method for the RDD prediction engine.
  4. Fourth and last, we will investigate if it is possible from a set of symptoms to predict whether a previously unseen disease is rare or not and, if so, how to prioritize its possible genetic and molecular determinants.