Dr. Narges Ahmidi

Creative research: Insight enables complex learning in machines

Understanding and insight – redundant skills when it comes to teaching machines to optimize medical standards in hospitals? Narges Ahmidi is convinced of the opposite: on this basis, the leader of the team ‘Machine Learning for Optimization of Patient Treatment’ at the Institute for Computational Biology (ICB) develops complex mathematical strategies for improved patient care.


Dr. Narges Ahmidi, Institute for Computational Biology (ICB). Source: Helmholtz Zentrum München

Even as a child, Dr. Narges Ahmidi read a lot. She filled the scenarios in the books with life in her mind and often adopted the role of the main character herself. The stories inspired her imagination and creativity. Today, she owes her ability to use computers to improve patient care in hospitals to her second talent – a pronounced understanding of mathematics. The synergy of these two assets is Ahmidi's magical formula. She had her personal formative experience when she was 16 years old: during a programming course at her school, she felt both of these deeply rooted aptitudes being satisfied in one fell swoop. "It was the first time I had ever used a computer and the first time I had studied creative math," she remembers. "It immediately took hold of me and has not let go to this day!"

Programming with imagination

In professional circles, it is well known that new mathematical strategies are not developed purely analytically, but, in addition to a keen sense of numbers, computational relationships and geometric patterns, it is mathematical imagination, in particular, that characterizes the intuitive and figurative thinking of mathematically talented people. Ahmidi is obviously one of them: Growing up in Iran, she studied Computer Engineering and Artificial Intelligence at Tehran Polytechnic University, was honored with the Iranian Presidential Young Scientists Award and the IEEE Student Award, and lectured at the University of Tehran. The versatile scientist acquired leadership qualities as a project manager and senior software engineer at the Institute for Telecommunications and Geographical Research in Tehran. She later earned a doctorate in Computer Science at Johns Hopkins University, Baltimore, USA. Here, she specialized in the fields of Medical Robotics as well as Computer Assisted Surgery and Machine Learning.

Today, Ahmidi still researches at Johns Hopkins University. With her team of fourteen, she collects patient data to subsequently allow machines to determine the best patient care strategy. Machine Learning is her mission. "A patient is routinely cared for before and after surgery, and much of this is routine even during the procedure," the engineer explains. Large volumes of statistical data, derived from these processes, are available to her. "I use them to identify which treatment guidelines prevent complications, that is, ensure that the patient can be discharged faster and more stably."

Trained Numbers

In reality, Ahmidi trains algorithms: she feeds data into the computer, i.e. situational descriptions as well as clinical events. The computer then interpolates these combinations and thus gets "a feel" for reality. After this training, the program can make its own predictions, for example to warn about complications before they occur, or assess the quality of treatment steps individually for each patient. "All of this without them being subjectively biased," explains the engineer. "And they are more precise the greater the data availability.”

Two terabytes of data

Her work benefits patients, doctors and hospitals: when their mathematical models calculate the best therapy, physicians automatically have an efficient and therefore cost-effective treatment plan, and hospital management knows in which areas meaningful investment is needed. In her dual function as leader of the research team ‘Machine Learning for Optimization of Patient Treatment’ at ICB, Helmholtz Zentrum München,  and the Johns Hopkins University team, she can now optimally put this project into practice: now Ahmidi develops and writes the programs that allow machines to calculate efficient treatment strategies together with Prof. Fabian Theis – the ICB Institute Director is an expert in the field of computational biology.

This work is based on more than two terabytes of real data from Baltimore. The new team leader therefore utilizes not only ICB expertise, but also supports its applications orientation. She says: "I aim to move digital healthcare in Germany forward."

The scientist from Tehran appreciates the expertise at ICB and also enjoys the open, intercultural climate. "The staff cooperate purposefully across cultural and professional boundaries," she says. Her imagination and empathy, which allow her to slip into different roles, also helps her here: "I find it exciting to adopt the way of thinking of both doctors and engineers, and find the interdisciplinary work very inspiring."