Home> News> #Bigdataforbloodcancer Blog: Artificial intelligence to decode Hematologic Malignancies

#Bigdataforbloodcancer Blog: Artificial intelligence to decode Hematologic Malignancies

May 27, 2021 20:29 - x 00, 0 - 00:00

Big Data Platform, BigData

Scientists are looking into the spectrum of data collected on the HARMONY Big Data Platform to find specific disease biomarkers in blood cancer, Hematologic Malignancies.


Gastone Castellani, Professor of Applied Physics and Biophysics at the University of Bologna, an expert in machine learning and neural networks of the HARMONY, explains how this project, the largest of its kind, can foster changes in clinical practice. The University of Bologne is one of the Partners of the pan-European HARMONY Alliance, including over 100 public and private organizations.



“The first challenge was to prepare a robust research facility. We had to establish the technical infrastructure and shape the security procedures for the secondary use of data. Having the right structure, a network of international experts, and data from over 35.000 patients at the present time, we have recently proceeded to the core phase – data analysis using artificial intelligence (AI). We are excited about the potential discoveries that lie ahead,” says Professor Gastone Castellani.

The scientists collaborating under the HARMONY Alliance hope to discover unknown relations between the data and thus develop new algorithms to be adapted in clinical practice. AI has many, so far untapped, opportunities in hematology. It can improve the precision of diagnosis, support risk stratification for instance for Acute Myeloid Leukemia (AML) and Myelodysplastic Syndromes (MDS), optimize treatment plans, and predict the response to different drugs. Read about the HARMONY blood cancer focus >


Correlations between data variables provide new insights

Medicine often lacks meaningful knowledge about the patient and the disease’s specifics. Although we can measure a variety of hematological parameters, we remain “data rich and information poor.” Therefore, we need Artificial Intelligence, which can analyze the large data sets available in Electronic Medical Records, including complex “omics” data, such as genomics, metabolomics, and epigenomics.

“If compiled and processed by machine learning or deep learning systems, data can reveal correlations that cannot be detected using manual methods,” highlights Professor Castellani. For example, having thousands of data sets on the HARMONY Big Data Platform, it’s possible to identify subgroups of patients with different features and compare their responses to therapy, the recurrence of the disease, and survival rates. “This approach – called unsupervised machine learning – is already showing promising results in our study regarding Acute Myeloid Leukemia and Myelodysplastic Syndrome,” says Castellani. The knowledge gained can be leveraged to develop algorithms – new tools for clinicians that every patient will benefit from.


The vision is timely diagnosis, personalized treatment and predictive medicine

New diagnostics methods with next-generation sequencing at the forefront lead to an exponential increase of available patient data. “The more data we have, the more accurate algorithms we can create.” Professor Castellani sees enormous potential in neural networks. These groups of algorithms can recognize relationships between vast amounts of data. In clinical practice, they can be applied to predict drug response, adjust chemotherapy, and choose the care plan that was most effective among patients with similar chemical and genomic features.

“One day, we will be able to create a digital twin, an in-silico replica of a human being. Imagine a collection of data that describes the patient. Algorithms developed using data collected on the HARMONY Big Data Platform could quickly perform calculations to support clinical decision-making.”
Professor Castellani mentions that the research he is involved in under the HARMONY Alliance may also enhance drug development or speed up clinical trials. “In the future, when treating patients with hematological cancers, clinicians will be able to choose personalized therapy options from a range of scenarios suggested by algorithms,” he adds.


Scientists design algorithms to answer open questions

The HARMONY Alliance has the intellectual and infrastructure capabilities to lead the scientific research regarding AI applications in hematology. Equally important is an innovative approach to the challenges that AI is facing. Researchers want to understand AI models’ decisions, to learn not only what an algorithm predicts but also why it makes a specific conclusion.

“No one can accomplish this mission alone. We are making progress due to the close cooperation among experts from data science, artificial intelligence, deep learning, medicine, hematology, genomics, physics, mathematics, and many more fields. Our international partners donate data that is for us the most valuable resource enabling and fostering our research,” concludes Professor Castellani.


The HARMONY Big Data Platform: central repository where the anonymous data donated by our partners and Associated Members is collected securely following all legal and ethical requirements, harmonized and then analyzed. Here emerge various data sets from different data providers – pharma, biobanks, hospitals, interventional, and non-interventional trials.


Related articles:


HARMONY (since 2017) and HARMONY PLUS (since 2020) have received funding from the Innovative Medicines Initiative 2 Joint Undertaking, under grant agreement No. 116026 for HARMONY and grant agreement No. 945406 for HARMONY PLUS. This Joint Undertaking receives support from the European Union’s Horizon 2020 Research and Innovation Programme and the European Federation of Pharmaceutical Industries and Associations (EFPIA).


Become involved! Connect with us and find out what we can offer you >


Receive the latest news. Click here to subscribe!