There are different kinds of software simulators of human physiology.
Some like BioGears aim at simulating a whole body, and how it evolves when it is harmed or when a substance is injected in it.
Others like BioDmet are intended for pharmacology research, and they study how a foreign substance is assimilated and propagates through organs.
* Simulation grain
Those software are also different in the simulation grain they provide. While some software simulate organs or even sometimes cells, others only make a rough simulation of the whole organ using for example the Glomerular Filtration Rate. Even if cells are simulated, the tissues they are in, are supposed to be homogeneous at organ scale, they offer no fine grained physiology simulation. No software will simulate tissues like Henle loop in kidney, or the retina’s epithelium. But most scientist teams need to work on such tiny parts of the body. Most of existing modeling software even ignore that there are two kidneys or two lungs, most modeling software simulate them only as a whole. The organ illness is taken in account with an overall parameter which degrades the model output, and not as a consequence of model’s input parameters and architecture. This may help to simulate accurately a in vivo model with some tinkering, but it does not help to understand why the illness exists in the first place.
* Software models should be based on biological concepts.
Most modeling tools use an approach based on mechanical properties of tissues, which is named ACME for “Absorption, Distribution, Metabolism, and Excretion”. One reason why it is very effective, is that good mathematical (ODE) and software tools (ODE solvers) exist. While effective until now, this approach ignores basic biological effects such as cells and tissues growth and depletion. It also ignores metabolic or signaling pathways, making this approach entirely useless for whole classes of biological phenomenas.
Other modeling tools are based on approaches even more foreign to physiology and biology, for example BioGears’ debuts were based on an electrical simulator (SPICE), so every BioGears model is still to be thought as electronic circuits.
* Integration in laboratory’s tool chain
No existing modeling software is really integrated in the laboratory tool chain. In vitro testing is quite useful in the characterization of some specific processes taking place inside the living organism. These processes can be integrated into the pharmacokinetic (PBPK) models. However this integration is quite challenging and often it is subcontracted to dedicated research services. This is particularly important when working with commercial Organ on Chips where the organ features and behaviors are well characterized but the underlying model is still unknown. As modeling tools are supposed to be a part of a tool chain between in vitro models and in vivo models, the software modelling tool should also create a standardized model that could be reused in other parts of the laboratory tool chain.
* Modeling software should enable to test multiple models.
Most software have physiological knowledge (models) embedded in code, only a few software load this kind of physiological information from human readable files. Not only this renders evolution of those software difficult and costly, but it hides their implicit model from the scientist and forbids her to substitute one model with another, in order to fine tune them to simulate accurately an organ on chip.
At the same time there are still no tools to create organ size models in computer biology exchange languages like SBML or CellML.
* Simulating accurately foreign pathogens and substances.
To solve some problems it is necessary to know how a pathogen will develop in human body, where it will likely thrive. Any human physiology simulation tool must include a good simulator of those pathogens in the human body.
* Reducing cost of drug discovery
In silico research in medicine is thought to have the potential to speed the rate of discovery while reducing the need for expensive lab work and clinical trials. One way to achieve this is by producing and screening drug candidates more effectively. A good modeling tool should automatically discover and propose a list of drug’s desired properties and drug candidates.
* Pre-clinical studies
Human beings weight and height are not standardized , more so ethnicity, family history and other factors create variations, for example in kidney modelling. A drug should work in all those cases and a drug candidate should be proposed not as a result of a statistical analysis, such as done in population modeling, but as a result of the model inner working. It should be fact based, not the result of a black box model, to help this drug to comply with regulatory bodies requirements and get required authorization.