Probiotic bacteria, particularly lactobacilli, play a key role in maintaining the balance of the gut microbiota and are used both in medicinal products for various therapeutic purposes and in dietary supplements and foods. To ensure their efficacy and stability during storage, freeze-drying is commonly employed. The thesis focuses on a systematic review of scientific studies addressing the freeze-drying of lactobacilli and includes an analysis of the potential application of artificial intelligence (AI) tools in the research process.
A systematic review of studies indexed in the PubMed database up to October 23, 2024, was conducted. In addition to the traditional manual review, a large language model (LLM) was used for data extraction, enabling the analysis of methodological aspects of the studies. Structured prompts were designed to systematically collect data on bacterial strains, freeze-drying parameters, protective substances used, microencapsulation, and storage conditions.
Out of 566 initial results, 272 relevant studies were selected. Most studies focused on individual strains, with Lactobacillus plantarum, L. acidophilus, and L. casei being the most frequently investigated species. The analysis of freeze-drying parameters showed that freezing was most often performed at –80 °C, while primary drying typically took place at –50 °C. The secondary drying temperature generally ranged between 20 and 25 °C, and the freeze-drying cycle most frequently lasted between 24 and 48 hours. The most often stabilizers used were sucrose and trehalose, with protein-based materials such as skimmed milk also frequently employed. Microencapsulation was used as an additional formulation step, with alginate and whey proteins being the predominant materials, and microcapsules the most common formulation type. Storage of probiotics was mostly performed at 4 °C for periods ranging from one to six months. The results demonstrated that AI already serves as a valuable support tool in the research process, particularly for the systematic analysis of a large number of sources.
|