High hydrostatic pressure sterilization: A resource optimization approach

High hydrostatic pressure (HHP) is a nonthermal sterilization method with strong potential for application in liquid and gel-like materials. Compared with conventional heat- or chemical-based approaches, it offers the advantage of preserving material integrity while reducing energy consumption. In this study, we combined machine learning (ML) with sequential experimentation to optimize the key parameters of HHP treatment (pressure, holding time, and number of cycles) and introduced a new output descriptor, termed effort, which integrates these variables into a single measure of process efficiency. The experiments demonstrated that complete inactivation of Escherichia coli could be achieved at 200 MPa for 10 min with four pressure cycles, providing the same level of sterilization as 300 MPa with a single cycle, but requiring approximately 25–30 % less energy. Energy analyses further revealed that increasing the pressure from 200 to 300 MPa raised energy consumption by about 70 %, whereas applying multiple cycles at lower pressures achieved an equivalent level of microbial reduction with substantially lower resource use. By integrating Random Forest modeling with Monte Carlo simulations, we identified parameter regions (≥200 MPa, 2–4 cycles, 5–10 min) that consistently combined substantial microbial inactivation with low effort. This approach reduced the number of experiments needed by more than half compared with traditional trial-and-error methods. In summary, the results demonstrate that ML/DoE-integrated optimization of HHP enables reliable bacterial inactivation under milder conditions and with significantly reduced energy demand. The proposed framework, therefore, provides a practical pathway toward more efficient and sustainable sterilization processes, particularly for sensitive biomaterials.