This study proposes a novel bio-inspired meta-heuristic algorithm, the Python Snake Optimization Algorithm (PySOA), that mimics the hunting behavior of the python snake. These reptiles are not poisonous, but they hunt their prey through ambushes. They can detect their prey using senses such as smell, eyesight, and infrared vision. The hunting mechanism consists of three major phases: searching for prey, scanning for prey, and attacking the prey. The searching-for-prey step contributes to exploration, while attacking prey is dedicated to exploitation, and scanning for attack enhances the balance between the two. The mathematical model of the method improves convergence precision and global search capability by capturing the behavioral dynamics of Pythons. PySOA’s performance was assessed on 23 classical benchmark functions, 29 CEC-2017 benchmark functions, 10 CEC-2019 composite functions, and three real-world engineering problems. The outcomes were confirmed by 14 popular meta-heuristic algorithms (MAs). With an average improvement of 39.2%, the PySOA outperformed the compared algorithms across all 62 test functions, achieving the best mean fitness rank in 43% of test cases. By successfully balancing exploration and exploitation, these findings demonstrate that PySOA is both resilient and competitive in addressing unimodal and multimodal optimization problems. The composite CEC-2019 test fitness functions demonstrated PySOA’s ability to explore and exploit simultaneously. The outcomes of the CEC-2017 benchmark tests show that PySOA has a shortcoming in local search exploration. Based on PySOA’s performance on real-world engineering problems, it is a practical algorithm for achieving optimal results and can be applied to real-world problems. The source code of the PySOA is publicly available at https://www.mathworks.com/matlabcentral/fileexchange/175654-a-novel-bio-inspired-python-snake-optimization-algorithm.

Safeguarding medical images against unauthorized access and alteration during storage and transmission is a critical challenge in modern telemedicine systems. This paper introduces a robust method to encrypt medical images in which the confusion stage is driven by a four-dimensional (4D) fractional-order chaotic system, and the diffusion process utilizes a symmetric matrix integrated with a one-dimensional (1D) chaotic map. The fractional 4D chaotic system reveals intricate dynamic behavior and is extremely sensitive to initial conditions, which enhances the confusion capability by thoroughly scrambling pixel positions. The symmetry matrix is combined with a generated chaotic sequence from a 1D chaotic map during the diffusion process that ensures strong pixel intensity diffusion and key dependence. Numerous experiments carried out on a variety of medical images confirm the outstanding performance of the suggested method. The suggested method features a key space exceeding 2¹⁰⁰, exhibiting significant robustness to brute-force attacks. It achieves unified average changing intensity (UACI) values above 33% and number of pixels change rate (NPCR) values exceeding 99.6%, confirms robustness to differential attacks, and successfully resists chosenplaintext and known-plaintext attacks. Additionally, the low pixel correlation and uniform histograms, along with average values of information entropy of 7.9973 and 7.9993 for 256×256 and 512×512 images, respectively, demonstrate strong resilience to statistical attacks. Furthermore, robust evaluations against cropping and noise attacks demonstrate the scheme’s practical security, highlighting its suitability for the safe storage and transmission of medical images in healthcare applications. Compared with related methods, the suggested method offers superior security performance.