A team of researchers led by Prof. Anubha Gupta from Indraprastha Institute of Information Technology (IIIT-Delhi) and Dr Ritu Gupta, Professor and Head, Laboratory Oncology Unit, All India Institute Of Medical Sciences (AIIMS, Delhi) undertook research with an objective to provide a robust system to test the cancer risk staging. 

Simply put, cancer staging is the process of finding out how much cancer is in a person's body and where it's located.

The team with the help of machine learning methods developed a new risk stratification system – Modified Risk Staging (MRS) for multiple myeloma (MM). Multiple myeloma, aka Kahler's disease, is a cancer of plasma cells. 

The plasma cells are a type of white blood cell in the bone marrow. With this condition called MM, a group of plasma cells becomes cancerous and multiplies. The disease can damage the bones, immune system, kidneys and red blood cell count.

This AI-enabled risk-staging system - MRS (a risk calculator) is done for newly diagnosed multiple myeloma (NDMM) patients. Prof. Anubha Gupta (IIIT-Delhi) mentioned, "This AI model was developed on a training dataset of NDMM patients and validated on two test datasets. A rigorous comparison of the proposed risk staging model with ISS and RISS was undertaken to check its efficacy on the predictions of progression-free survival (PFS) and overall survival (OS)."

The model takes in six easy-to-acquire laboratory parameters, namely:

• Age (years)     
• Albumin (g/dL)  
• B2M (mg/L)   
• Calcium (mg/dL)          
• eGFR (ml/min)
• Hemoglobin (g/dL)

Prof. Ritu Gupta (AIIMS) stated "It is an efficient and readily employable risk prognostication method that is beneficial for the settings where genomics tests cannot be performed owing to geographical or economical constraints. Hence, the research paves the way to predict the risk factor in a more efficient and cost-effective manner that can be readily available to the masses at large." 

Also, the study recommends training of machine learning models on larger datasets because that can provide efficient upfront prognostication that may be useful in the selection of therapy of appropriate intensity, especially in high-risk MM patients. Further, the performance on the MMRF dataset indicates that there may be nuances with respect to ethnicity and race.