Supporting personalised and preventive medicine with the first Acute Myeloid Leukaemia Digital Twin system
Frank Emmert-Streib and Olli Yli-Harja are building a prototype of the AML Digital Twin system with their research team. AML-DT is an interactive system to be used by doctors with their patients. In the system, the patient’s clinical and molecular data will be collected from bone marrow aspirates and peripheral blood and it will be used to instantiate a digital twin model of the patient’s disease.
The work is conducted jointly with clinician-scientists and computational researchers Emmert-Streib and Yli-Harja at Tampere University, Ilya Shmulevich at the Institute for Systems Biology in the USA, Caroline Heckman and Mika Kontro at the Institute for Molecular Medicine Finland at the University of Helsinki University with Kontro also affiliated with Helsinki University Hospital.
Preventive medicine vs. reactive medicine
The digital twin can be used to maintain health by using its personalised predictive capacity, which allows early intervention to avoid disease. The digital twin can predict transitions within the course of disease development.
“In diabetes, for example, some important transitions occur between pre-diabetes and diabetes and between diabetes and kidney disease. These and other negative transitions can be predicted with the digital twin system and potentially avoided with timely intervention”, Yli-Harja concludes.
In case of a metabolic disease, such as diabetes, a successful intervention depends heavily on the patient’s participation, actions, and motivation.
“With the help of the digital twin dashboard the clinician can involve the patient in the decision-making process and strengthen the patient’s participation. In addition to benefitting AML patients, our AML-DT prototype holds great promise for improving the way diseases are treated in general,” Yli-Harja continues.
Digital twin supports stronger patient participation in treatment
The digital twin will enable the doctor and patient to explore personalised model-based predictions of drug response.
“It can be done by using measurable outcomes in the light of background knowledge obtained from publicly available molecular data from Acute Myeloid Leukaemia (AML) patients. Importantly, the AML-DT system will continuously improve by learning from the experience of patients and their digital twin models,” Emmert-Streib explains.
The digital twin supports the patients to participate more in their treatments, for example, in making health decisions. Yli-Harja says that this will lead to a stronger commitment of the patient.
“Integration of knowledge from databases combined with personalised models allows us to obtain personalised predictions for the patients. This will enhance conventional approaches that are often based on patient cohorts treated in the same way. Hence, a digital twin system embraces the ideas of personalised medicine by having fine-tuned models for individual patients,” Yli-Harja points out.
Conventional chemotherapy has reached its limits in treating AML
AML is the most common leukaemia in adults. Many patients who receive intensive chemotherapy and healthy blood-forming cells from a donor eventually relapse. Therapeutic results do not provide promising results for relapsed or refractory and older patients who are often unfit for intensive therapies.
Several new therapy options have emerged with the potential to improve treatment outcomes. However, they suffer from the heterogeneity of responses or lack methods to identify patients who could benefit from such therapies. AML-DT is meant to meet the needs of the patient more comprehensively.
“The proposed AML-DT prototype system visualises patient history, makes predictions of disease progression and drug response which can be validated in clinical trials, and gives clinicians and patients a way to make better treatment plans by providing much needed foresight,” Emmert-Streib says.
In the study, Emmert-Streib and Yli-Harja will integrate detailed specialised knowledge graphs with multiscale dynamic models of the malignancy. It means that general information about the disease is integrated with molecular information to obtain a personalised model. In this way, they can provide a foundation for the development of the first AML Digital Twin.
“In addition to this, we expect that our results can also be expanded to other diseases requiring decision support systems. From a more fundamental data science and AI (artificial intelligence) perspective, the project has the potential to serve as a proof-of-concept for a simulation-based inference which can also be of relevance in many different application domains, including industrial engineering and climate research”, Emmert-Streib adds.
The research project is a part of the AKA-NIH Partnership Programme launched in November 2020. It is an initiative involving funding agencies from Finland and the United States. In Finland, the partner agency is the Academy of Finland (AKA) while the partner agency is the National Institutes of Health (NIH) is the partner in the US. The overall goal of the AKA-NIH Partnership Programme is to increase research collaboration between Finland and the US.
Enquiries:
Professor Frank Emmert-Streib
frank.emmert-streib [at] tuni.fi (frank[dot]emmert-streib[at]tuni[dot]fi)
+358503015353
Professor Olli Yli-Harja
olli.yli-harja [at] tuni.fi (olli[dot]yli-harja[at]tuni[dot]fi)
+358408490778
