Expensive experiments
PIML can extract more value from limited fermentation, cell-culture, assay and imaging data by combining them with established mechanisms.
Engineering Research Community
Biotechnology & Physics-Informed Machine Learning
Connecting biological systems, engineering principles and artificial intelligence
Biotechnology combines biology, chemistry, engineering, mathematics and computation. Its experiments can be expensive, slow and difficult to repeat, while living systems are nonlinear, variable, multiscale and only partially observed.
Physics-Informed Machine Learning (PIML) helps researchers combine experimental observations with mass balances, reaction and enzyme kinetics, cell-growth models, transport phenomena, biochemical pathways and process constraints. The aim is not to replace biology with equations, but to create useful grey-box models that learn what remains uncertain.
This page gives students and researchers a practical starting point: ten high-value research areas, project ideas by academic level, selected publications, a validation pathway and direct access to the PIMLS community.
This Biotechnology guide covers Physics-Informed Neural Networks (PINNs), physics-guided machine learning, scientific machine learning, neural operators, hybrid models and engineering digital twins. Explore the research and project pathways below, then join the Physics-Informed Machine Learning Society to connect with the international PIMLS community.
The central ideaBiological knowledge + physical and chemical laws + experimental data + machine learning
10focused research areas
3academic project pathways
6selected publications
Biweeklymember research meeting
Why this combination matters
Why Biotechnology Needs Physics-Informed Learning
Use available scientific knowledge to make limited data more useful, transparent and testable.
Partial biological knowledge
Known pathways, kinetics and balances can be retained while machine learning estimates missing rates, closures or hidden states.
Multiscale systems
Biotechnology connects molecular, cellular, tissue, organ and bioreactor scales; structured models help carry information between them.
Trustworthy decisions
Physical and biological checks, uncertainty and external validation make predictions easier to interrogate before research or process use.
Ten focused directions
Major Biotechnology PIML Research Areas
Each card connects a meaningful Biotechnology question with suitable scientific knowledge, modelling choices and evidence needed to test it.
01
Bioprocess Engineering
Hybrid models for biomass, substrate and product prediction, feeding strategies, process monitoring and scale-up.
Model and evidenceMass and component balances; reaction kinetics; process sensors
02
Fermentation Technology
Physics-informed fermentation models for yield, productivity, endpoint prediction, oxygen limitation and optimal operation.
Model and evidenceMonod-type kinetics; heat and mass transfer; batch data
03
Bioreactor & Cell Culture
Digital twins for mixing, oxygen transfer, viable cell density, nutrients, metabolites and changing operating conditions.
Model and evidenceCFD or compartment models; cell kinetics; PAT measurements
04
Biopharmaceutical Manufacturing
Mechanistic ML for upstream cultivation and downstream filtration, chromatography, purification and real-time control.
Model and evidenceProcess models; quality attributes; validated assays
05
Enzyme & Metabolic Engineering
Learn kinetic parameters, reaction rates, metabolic fluxes and strain behaviour while respecting stoichiometry and enzyme mechanisms.
Model and evidenceMichaelis–Menten kinetics; stoichiometry; omics and perturbations
06
Systems & Synthetic Biology
Model cell signalling, gene circuits and regulatory networks and discover unknown dynamics without discarding known biology.
Model and evidenceReaction networks; neural ODEs; perturbation experiments
07
Tissue Engineering & Mechanobiology
Couple tissue growth, scaffold mechanics, nutrient transport, cell migration and mechanical signalling.
Model and evidenceSolid/fluid mechanics; diffusion; imaging and biomechanical tests
08
Biosensors, Biofluids & Drug Delivery
Reconstruct flows and analyte states, calibrate biosensors and model diffusion, release and transport through tissue.
Model and evidenceSensor physics; fluid mechanics; diffusion and compartment models
09
Environmental & Industrial Biotechnology
Support wastewater treatment, anaerobic digestion, bioremediation, biofuels, food biotechnology and biocatalysis.
Model and evidenceSpecies balances; microbial kinetics; environmental measurements
10
Digital Twins, Inverse Problems & UQ
Estimate hidden states and biological parameters, update real-time twins and report uncertainty across unseen conditions.
Model and evidenceMechanistic simulators; data assimilation; uncertainty calibration
Choose the right research depth
Projects for Every Academic Stage
Start with a scope that matches your time, mathematical background, experimental access and expected research contribution.
Project pathway 1
B.E./B.Tech
Learn the foundations with a bounded, measurable system.
- Fermentation yield prediction with a Monod-kinetic penalty
- Enzyme-kinetics parameter estimation from laboratory data
- Physics-aware biosensor calibration and drift detection
- Simple batch-bioreactor mass-balance neural model
- Drug-release diffusion model with sparse measurements
- Wastewater microbial-growth prediction with balance checks
Expected outcome
A reproducible notebook, clear baseline, small experimental or published dataset and a concise validation report.
Project pathway 2
M.Tech
Combine a mechanistic model, substantial data and rigorous comparison.
- Hybrid bioreactor model for oxygen transfer and cell growth
- Bioprocess soft sensor for biomass, substrate or product state
- Mechanistic ML for chromatography or filtration
- Metabolic-flux estimation using stoichiometric constraints
- Tissue nutrient-transport inverse problem
- Uncertainty-aware fermentation or cultivation digital twin
Expected outcome
A thesis-quality study with mechanistic and data-only baselines, held-out operating conditions, ablation and uncertainty.
Project pathway 3
Ph.D.
Address a publishable methodological or multiscale research gap.
- Neural operators for families of bioreactor conditions
- Multiscale cell–tissue–reactor scientific machine learning
- Physics-informed neural ODE/SDE models for biological dynamics
- Hybrid discovery of unknown biochemical mechanisms
- Real-time biopharmaceutical digital twins and safe control
- Transferable uncertainty-aware models across laboratories or scales
Expected outcome
New methodology or validated scientific insight, multi-regime evidence, uncertainty, reproducible software and journal publications.
From idea to evidence
A Strong PIML Project Workflow
01
Define
Choose one Biotechnology question and a measurable engineering output.
02
Model
State the governing relationships, constraints or validated domain knowledge you will retain.
03
Compare
Build mechanistic and data-only baselines before the hybrid model.
04
Validate
Hold out experiments, conditions, assets, sites or regimes at the deployment level.
05
Publish
Report uncertainty, ablation, limitations, data lineage and reproducible code.
Read before you modelDiscuss Your Literature
Selected Publications and Why They Matter
Use this focused reading list to understand the general PIML framework, direct Biotechnology evidence and suitable hybrid modelling methods.
Literature review advice
Do not list papers only. Compare the engineering question, incorporated knowledge, data, split strategy, baselines, uncertainty and evidence level.
A broad review of physics-informed learning across biomedical and biological systems, including PINNs, neural differential equations and neural operators. Use it to map methods to biofluids, biosolids, mechanobiology, pharmacokinetics, signalling and imaging.
How to use this paper: Best starting point for a biotechnology or biomedical literature review.
Read publication or record A directly relevant bioprocess example combining cultivation data with biological growth knowledge to improve prediction across changing process conditions.
How to use this paper: Useful for bioreactor, cell-culture, fermentation and biomanufacturing projects.
Read publication or record A recent perspective covering important PIML methods and biomedical applications while highlighting open challenges in evidence, interpretability and translation.
How to use this paper: Useful for selecting a contemporary research gap and understanding limitations.
Read publication or record The foundational review explains how data-driven models can incorporate governing equations, symmetries, constraints and numerical simulations across scientific domains.
How to use this paper: Cite this for the general PIML framework and method taxonomy.
Read publication or record The seminal PINN formulation uses differential-equation residuals and data to solve forward and inverse problems. Biotechnology studies should also discuss PINN optimization and validation limitations.
How to use this paper: A methodological foundation for diffusion, transport, reaction and parameter-estimation projects.
Read publication or record Universal differential equations provide a flexible pattern for retaining a mechanistic ODE/PDE model while learning an uncertain component from data.
How to use this paper: Particularly relevant to grey-box bioprocess, pathway and physiological models.
Read publication or record Build an interdisciplinary team
Where Biotechnology Can Collaborate
Computer Science
Scientific ML, neural operators, optimization, trustworthy AI and reproducible software.
Chemical Engineering
Bioreactors, transport, reaction engineering, separations and process control.
Mechanical Engineering
Biofluids, tissue mechanics, mechanobiology, mixing and thermal systems.
Electrical & Electronics
Biosensors, instrumentation, signal processing, embedded monitoring and control.
Before you begin
Frequently Asked Research Questions
These answers help students avoid common scope, terminology and validation mistakes.
Still have a question?
Use the biweekly meeting form for research guidance.
Request accessNo. Useful prior knowledge may be an ODE, mass balance, stoichiometric relation, enzyme or growth kinetic model, diffusion law, compartment model, pathway structure or validated simulator. State exactly what knowledge is used and where it enters the learning process.
Choose one measurable output, a small defensible mechanistic model and a dataset with clear units and conditions. Establish mechanistic and data-only baselines before adding a hybrid model.
A meaningful biological question, justified prior knowledge, held-out conditions at the correct experimental level, strong baselines, ablation, uncertainty, reproducible code and honest limitations.
Simulation can broaden coverage, but simulation-only testing cannot establish real biological or process accuracy. Use independent assays or experiments whenever the claim concerns a real system.
Submit the biweekly members meeting form. The Society can use the meeting to understand your branch, project level, model, data and collaboration needs.
Take the next step
Bring your Biotechnology research idea to PIMLS
Join the biweekly members meeting for project guidance, collaboration and publication planning—or contact the Society directly.
Meeting participation is requested through the Google form. Complete it carefully so the Society can understand your research interest.