Insight into airflow in human lungs can provide vital information about the health of the respiratory system and its treatment. In particular, the efficiency of inhaled drug targeting can be highly improved using Computational Fluid Dynamics (CFD) techniques. Drug delivery optimization to assure the deposition of inhaled drug at the target location in the human airways paves way for personalized inhaled medicine. One Simulations has been working in collaboration with research and industrial partners to make it a reality.
According to the WHO, a total of 338 million people suffer from asthma worldwide. The prevalence of asthma in the European Union (EU) is 8.2% in adults and 9.4% in children which amounts to around 40 million people in the EU alone. Most common medication for the severe and mild asthma cases are inhalers that deliver drug to the airways either as instant relievers in case of an asthma attack or as daily inhalers that gradually calm down the inflammation and sensitivity in the airways. However, these inhaler drug particles are carried by the air flowing in the human airways and hence makes it difficult to control drug delivery to specific locations in the airways where the inflammation occurs. The existing method relies on the likelihood that some of the drug particles are delivered to the required area. However, using CFD this can be vastly improved by choosing the right kind of drug particle that deposits with higher precision on a specific target area, thus making the drug delivery more efficient.
To achieve such an objective, it is imperative to understand the effect of geometrical parameters in human lungs on the airflow and therefore drug deposition, which would require an extensive parametric study. This would mean gathering lung geometries for a large number of patients to execute CFD analysis, which is a tedious and time taking task. Moreover, finding a CT scan geometry, cleaning it to make it suitable for a CFD mesh and executing the CFD simulations itself is a time consuming process. An easier and a much more efficient way is the use of mesh morphing tools to morph (modify) a (single) base mesh and hence simulate the effect of various geometrical parameters on the air flow in human lungs. We are working in partnership with mesh morphing experts to make this a possibility. An example is shown in figure below where figure (a) shows typical lung geometry circled red in the esophagus region , (b) zoom-in view of the minimum cross-sectional area of the esophagus region marking the width of constriction as D, (c) minimum cross-sectional area reduced to 80% by morphing (constriction width 0.8D) the mesh using RBF Morph tool while keeping the rest of the geometry and mesh same.
As a part of the development, the future plan includes use of Reduced Order Models (ROMs) to mimic the airflow in human lungs. This technique uses machine learning and artificial intelligence algorithms to create data driven models (ROMs) which can mimic the physical CFD model. If an extensive set of data is provided to such a tool, it would be able to predict the airflow in most human airway geometries without having to execute a CFD analysis for every individual patient, saving a lot of time and costs.
The final result of the initiative would be a Digital Twin (DT) of the human airways. A technique based on compression methods able to drastically reduce the complexity of large-scale numerical simulations while maintaining a good level of detail. The synergic use of ROM methods, artificial intelligence, machine learning and software analytics, is at the core of the DT, a responsive digital replica of the complex original system, that can be explored by tweaking its parameters and inspecting in real time their influence. Generating a DT of the human airways, paves the way for personalized medicine in customized drug delivery.
 Kenjereš S, Tjin JL. 2017 Numerical simulations of targeted delivery of magnetic drug aerosols in the human upper and central respiratory system: a validation study .R. Soc. open sci. 4: 170873. http://dx.doi.org/10.1098/rsos