L-carnitine is a natural amino acid that is often used to lose weight. It plays a vital role in energy production by transferring fatty acids into mitochondrial cells. The mitochondria act as intracellular motors and burn these fats to provide the energy needed by the body. In addition to the major advantage of facilitating weight loss, research has shown that it has the benefits of improving brain activity and preventing the decline in brain activity in patients with Alzheimer's disease, lowering blood pressure, improving inflammation and heart disease and reducing cardiac arrest, improving body function in endurance exercises, reducing symptoms and complications associated with type 2 diabetes, and maturation and increasing sperm fertility. The body can produce L-carnitine virtually through the lysine and methionine amino acids. In addition, a person can receive L-carnitine from external sources, such as animal and plant products, as well as supplements. Because some people may have a deficiency of this substance that can be caused by genetics, certain medications (valproic acid for seizure), or the effects of some therapies (hemodialysis for kidney disease), therefore, the supply of this material has always been of interest from external sources. However, animal sources contain significant amounts of L-carnitine, but research has shown that among plant sources, mushrooms and, in particular, oyster mushrooms contain high and comparable amounts of L-carnitine. Supercritical extraction, in particular extraction with carbon dioxide, has been considered in recent decades due to its advantages such as non-flammability, non-toxicity, high purity, low prices, lack of solvent residue, higher selectivity, higher speed, and higher yields. In this study, extraction of L-carnitine from oyster mushroom using supercritical carbon dioxide and methanol as co-solvent was performed on a laboratory scale. The response surface method was used to analyze the results of extraction. For this purpose, the effect of parameters such as pressure (10-30 MPa), temperature (40-80 °C), flow rate (0.5-2.5 ml/min), and dynamic time (40-120 min) on the rate Recovery of L-carnitine was investigated. The response level analysis showed that the laboratory data was well fitted with a second-order polynomial model, with a determination coefficient of 93.05% and a correction coefficient of 86.98%. In addition, it was found that linear and second order semantics of all parameters, together with the interaction between temperature and pressure, have a very important effect on the recovery rate of L-carnitine. Also, the investigations showed that the optimal values of extraction parameters in the experimental range were 40 °C, 19.5 MPa, 2.14 ml/min flow rate and 120 min extraction time, under which the recovery rate of L-carnitine was equal to 55.28% is predicted, which with performing 3 tests at the optimal point, the relative error was obtained at ±0.4%. In addition, in the present study, the extraction process is modeled using a mathematical model. In the model, the effect of axial dispersion in the fluid phase, the film mass transfer resistance, and the hole diffusion in the particles is considered. The model parameters include effective diffusion coefficient, film mass transfer coefficient, axial dispersion, and distribution coefficient. The first three parameters are calculated by empirical relationships and the distribution coefficient, which is used to express the equilibrium state between the solid phase and the fluid inside the cavities, has been calculated through thermodynamic and genetic algorithms, in which the absolute average relative deviation between the model results and the experimental data minimize. The values of absolute average relative deviation obtained from thermodynamic and genetic methods and the coefficient of determination of the mathematical model in this study were less than 14%, 4% and 95.83% respectively, which indicates the compatibility and proper accuracy of this model. Finally, the process was experimentally modeled using artificial intelligence techniques such as artificial neural network and adaptive neuro-fuzzy inference system, so that the accuracy of these models based on their determination coefficient was 98.18% and 94.26% respectively which suggests that the experimental modeling with the artificial neural network is stronger than other modeling approaches. Keywords Supercritical extraction, co-Solvent, L-carnitine, Oyster mushroom, Carbon dioxide, Methanol, Response surface methodology, Mathematical modeling, Equations of state, Genetic algorithm, Artificial neural network, Adaptive neuro-fuzzy inference system