Machine Learning Aiding Management of Rare Diseases

Effective treatments are available for only about 5 per cent of Rare Diseases. Pharma has traditionally been hesitant to invest in this space as the development and commercialization of rare diseases is a challenging area for any company

The US FDA defines a rare disease as a disease that affects less than 200,000 patients in the US. There are about 7,000 Rare Diseases, and according to a report, 300 MM people live with a rare disease worldwide. 72 per cent of rare diseases are genetic, and 70 per cent of those rare genetic diseases start in childhood. 

The figures are astonishing, and more so, given the limited understanding of many rare diseases, patients struggle with getting the proper diagnosis and treatment. Also, effective treatments are available for only about 5 per cent of Rare Diseases. Pharma has traditionally been hesitant to invest in this space as the development and commercialization of rare diseases is a challenging area for any company.  

Further, the economics and the limited disease awareness make investments in the space more challenging to justify. 

However, lately, governments have been incentivizing the development and commercialization of treatments for Rare Diseases through various incentives and grants. These include: 

Extended Market exclusivity,  

Tax credits, Grants for drug development,  

Accelerated approval pathways (e.g., FDA Breakthrough Therapy designation)  

Reduced statistical burdens for clinical development  

Because of these concerted efforts, we are seeing a growth in rare disease research and increasing approvals for rare disease drugs. The positive trend is expected to continue for the foreseeable future.   

Challenges in Identification and Management 

Rare disease diagnosis is filled with several challenges and roadblocks. If we look at the statistics, it takes an average of 7.6 years for rare disease patients to get an accurate diagnosis. Patients typically visit several physicians before receiving an accurate diagnosis; chances are that the patient would receive 2-3 misdiagnoses before the correct diagnosis is determined.  

In Europe, disorders afflicting less than five patients per 10,000 are deemed rare. Yet, rare disorders are rare only when considered individually. Since more than 7,000 recorded rare diseases exist, the collective health burden remains high globally. Recent reports estimate a population prevalence of 3.5 to 5.9 per cent at the least. Yet, the actual prevalence is probably more since epidemiological data for many rare disorders remains scarce.  

There are many challenges that are unique to rare diseases. 

It starts with tackling awareness that could be quite low given the small, affected population. The patient advocacy group can play a critical role here.  

The diagnostic journey is very challenging and painful. During this time, patients go from one physician to another, go through several misdiagnoses and try to find a physician with the necessary expertise to provide a correct diagnosis. There is a lot of heartbreak and frustration during this journey, and, in some cases, it may be too late for certain patients because these diseases can be fatal. 

Thirdly, it is around Treatment Choice. The diseases and treatment options are extremely complex. They require a lot of scientific and clinical messaging to ensure the benefits of the available treatment are understood by the physicians as well as by the patients. So education is extremely critical. 

Access to therapy can be a very cumbersome process. These are expensive treatments and involve a significant amount of burden in navigating the insurance and managed care processes, and patients who are suffering need a lot of support to gain access.  

Finally, it is Patient Management. Patient support programs are critical for ongoing education and support. It is a rather long and complex cycle from disease identification to diagnosis to treatment. Patient support services will be essential in ensuring strong compliance and adherence to treatment. 

This diagnostic odyssey of patients makes rare disease management extremely difficult. As a result, novel tech-enabled solutions are required to support the decision-making process through automated and quantitative tools. In this situation, Machine Learning (ML) provides a plethora of powerful inference methods. However, matching health disorders with advanced statistical techniques leads to technological, methodological, and ethical issues.  

The Potential for using Machine Learning (ML) 

Fortunately, ML and data science advances have presented new possibilities for helping diagnose and treat these ailments. ML algorithms can be deployed to pinpoint patterns in large volumes of patient data to discover potential treatments[4]. ML can also help detect early warning signs of rare disorders while facilitating early intervention and timely treatment to prevent full disease onset.   

Besides, ML may be used to analyze the effectiveness of existing therapies, helping to optimize them while reducing side effects. Also, swiftly analyzing immense data sets from clinical trials and other sources can ensure faster, more accurate diagnosis and treatment.    

AI (Artificial Intelligence) and ML (Machine Learning) can revolutionize diagnosing and treating rare disorders. Digital tools and algorithms can help clinicians dig deeper into patient data, discovering correlations and patterns that are impossible to find through traditional means. By understanding the complexities of each patient, gaps in their needs can be identified, paving the way for personalized care.  

With AI/ML, unprecedented insights can be gained into the causes, symptoms, and treatment modalities. These tools and algorithms can also be leveraged to discover prospective drug targets and develop new therapies that help improve affected lives.   

Addressing Limitations and Ethical Considerations 

Nonetheless, one should realize that ML also has some limitations regarding rare diseases. For instance, ML algorithms require massive amounts of data to be trained and to discover specific patterns. Yet, since such disorders are rare, limited data exists that could be used for ML. This could lead to inaccurate results as the algorithms may not be able to detect patterns with high accuracy, considering the less-than-ideal robustness of the data.  

Additionally, even if ML detects patterns, it may need help to fully uncover solutions to these problems. Accordingly, human intervention is required to decode the complexities of rare disorders and provide proper solutions. Flaws in data pattern recognition can also be a hindrance since the algorithms may pinpoint a false pattern, given the limited data.   

Another concern in recent years has been around ethical considerations in using ML to leverage patient data. As new investments are made in AI and data analytics, the ethical implications of how the collected patient data is used must be considered.  

A code of ethics may need to be established for entities using ML for rare diseases, ensuring it is only used to benefit patients and not for commercial purposes. When collecting and using such patient information, data privacy must be paramount.  

Rare diseases require a deeply focused customer-centric support model that can guide the patients through disease awareness, to providing education and treatment support, and establishing trust for achieving successful patient outcomes and delivering improved quality of life 


Tags assigned to this article:


Around The World