Artificial Intelligence (AI), including Machine Learning (ML), can be used to solve various issues of information processing, including pattern recognition, classification, clustering, dimensionality reduction, image recognition, natural language processing, and predictive analysis [1]. AI-based algorithms can be applied to solve complex problems ^[2][3][4][2,3,4]. Recent algorithm development enables its application in many areas of everyday life, such as industry, medicine, and nanomedicine; including nanomaterials with magnetic properties [5]. Consequently, a new opportunity to predict drug influence and responsiveness based on retrospective databases became available [6]. It may contribute to the development of optimized healthcare [7].

An important direction in developing medicine is to provide an effective method of dealing with various neoplastic diseases. The heterogeneous nature of tumors contributes to the problems in selecting effective treatment mechanisms. It is crucial to deliver drugs directly to the tumor core, the area most active in proliferation but less vascularized and hypoxic. Thus, the critical challenge in choosing the optimal therapy is determining the synergy of the drug depending on its dose, administration timing, and current treatment process. The latest development in nanotechnology enables the design of nanocarriers for targeted drug delivery, improving medicine release and beating cancer cells. In turn, manufacturing the nanoparticles, which can be loaded with drugs or other agents (stabilizers, compounds for diagnostics), is a time and financial outlays-consuming process. Thus, the AI-based prediction of the effect of nanoparticles with drugs on living tissues enables the development of targeted nanomedicine ^[8][9][8,9].

Since each subject is different, and drug synergy gives a different output in an individual case, transforming Artificial Intelligence (AI) to nanomedicine enables the analysis of large data sets and the effective selection of the optimal therapy ^[10][11][51,52]. It is essential in cancer therapy, particularly in the application of magnetic hyperthermia, to predict the optimal parameters of the process. AI includes various algorithms; rein thisearchers paper, we reviewed the existing solutions in the area of research, which involve the use of magnetic hyperthermia, taking into account their effectiveness, type and size of data sets, input and output parameters, and application fields. In ^[12][59], ANN was applied to predict the size of AgNO₃ particles. It turned out that the most sensitive parameters are both AgNO₃ concentration and reaction temperature. As the AgNO₃ suspension has no relation with magnetic hyperthermia, the literature shows the successful use of ANN in the prediction of particular properties of nanomaterials. In ^[13][60], the ANN was proposed to predict the shape and size of TiO₂ nanoparticles. In Table 1, the algorithms for the evaluation of the nanoparticle size were compared. It turned out that neural networks, in particular networks based on multilayer perceptrons, enable the prediction of the size of nanoparticles with high accuracy, i.e., 0.97 based on the experimental data.

^17][64]. It turned out that AI-based prediction can substantially increase the relative thermal conductivity of nanofluids. In Table 2, the algorithms for the evaluation of the thermal conductivity of the nanoparticle were compared.

The AI-based analysis of thermal conductivity, taking into account different shapes of nanoparticles (i.e., spherical, ellipsoidal, clubbed, and sheet), has been made in ^[

The following research [5] described the ML application to the prediction of power losses of magnetic particles, which is an important issue in drug targeting. This researchtudy has been done on a limited amount of data containing simulated particles’ simulated properties, showing the proposed approach’s colossal potential. In Table 5, the algorithm’s performance in predicting power losses of magnetic particles has been made. The evaluation of the biological and mechanical behavior of the potential candidate for nanomaterials, i.e., silicates bioceramics-magnetite bio-nanocomposites, which can be applied to the magnetic hypothermia based on ANN has been made in ^[26][73]. In turn, the results presented by ^[24][71] show that ANN has better accuracy than genetic algorithms (GA) in predicting Young’s modulus and ultimate tensile strength of nanocomposites, particularly polyethylene composites with multiple nanoparticles.

^[37][38][39][83,84,85]. Thus, AI-based algorithms can be applied to predict removal efficiency. In ^[39][85] the application of the Artificial Neural Network and adaptive neuro-fuzzy inference system for the prediction of the chromium removal efficiency was shown. In Table 7, the comparison of algorithms for the evaluation of the removal efficiency is presented. It turned out that the combination of the Artificial Neural Network with an adaptive neuro-fuzzy inference system provides higher prediction efficiency.

One of the most important properties of nanoparticles is neurotoxicity ^[18][65]. In ^[20][67], the classification model for evaluation neurotoxicity based on Random Forest was proposed while in ^[21][68], the evaluation of antibacterial capacity using different AI-based algorithms was shown. In turn, ^[22][69] applied ML and perturbation theory to evaluate the toxicity of nanoparticles. In Table 3 the comparison of the algorithms for the prediction of the neurotoxicity of the nanoparticle was compiled. So far, the ANN can also be used to predict specific parameters for magnetic nanoparticles, see Table 4.

In ^[27][74], the application of the genetic algorithm for the optimization of agar nanospheres, which were used in the manufacturing process of drug loading, was proposed. The optimization problem was formulated to minimize the size of the particle, release efficiency, and PDI, as well as to maximize the absolute value of zeta potential and loading efficiency. It turned out that a genetics algorithm could successfully predict the parameters of Bupropion hydrochloride-loaded agar nanospheres. In turn, ^[28][75] the closed-loop optimization of the release process of the Poly (lactic-co-glycolic acid) (PLGA) biodegradable particles with ANN and genetic algorithms was described. As input data, particle size, and initial burst percent at the desired levels were chosen. It was postulated that the proposed algorithm can predict drug delivery.

In ^[29][76], the ANN was used to predict the optimal composition of two-dimensional graphene–Fe₃O₄ nanohybrids, which are dedicated to magnetic hyperthermia. It enables the prediction of the composition of the optimal nanohybrid, which can be applied to magnetic hypothermia in low dosage. The optimization based on multilayer perceptron neural networks of the experimental conditions of nanoparticles was described in ^[30][77]. The influence on nanoparticle characteristics factors like environmental conditions and type of precipitating agent was investigated. In turn, the mathematical framework for the magnetic drug delivery taking into account the ferrofluid flow was shown in ^[31][17].

Another critical issue in the manufacturing of nanoparticles is the synthesis process ^[32][33][78,79]. It should maintain precisely controlled characteristics. Since the synthesis of nanoparticles is a long-term and cost-consuming process due to the involvement of multiple chemical substances, the AI-based algorithm provides the opportunity to develop efficient experimental protocols. The following research describes the application of AI to the synthesis of semiconductor, metal, carbon-based and polymeric nanoparticles ^[34][80]. In ^[25][72], based on ultraviolet-visible (UV-vis) and PL spectrum data, the prediction of the optimal parameters of the synthesis of combinatorial CdSe nanoparticles was proposed. Thus, the heuristic and Bayesian optimization can be applied to the evaluation of the synthesis of the nanoparticles. Such an example is far from the magnetic hyperthermia application, while AI support can improve the experimental work also in the magnetic nanoparticles and magnetic hyperthermia field. In ^[35][81] the genetics algorithm particle swarm optimization (PSO) was used to predict the magnetic field generation. In ^[35][81], GA was used to optimize the Specific Absorption Rate in the case of hyperthermia treatment of the human head.

Recently, attempts were made to apply AI-based algorithms in the research of hydrogels. In reference ^[36][82], the Artificial Neural Network and Least Square Support Vector Machine were used to evaluate the swelling degree in the hydrogel, namely poly(NIPAAm-co-AAc) IPN. It turned out that Artificial Intelligence-based algorithms can, successfully and with high accuracy, predict the influence of pH and temperature on hydrogel deswelling behaviors. At the same time, the ANN model has higher computational efficiencies than the LS-SVM approach while maintaining this similar accuracy. Thus, in ^[37][83], ANN was used to evaluate the deswelling and heating behavior of the field-sensitive hydrogels, like poly(NIPAAm-co-VSA)/Fe₃O₄ IPN. The comparison of the algorithms for predicting deswelling behaviors is made in Table 6. It turned out that ANN achieved the highest efficiency in predicting deswelling degrees.

Magnetic nanoparticles are also used to remove various types of substances. The efficiency of the approach is strictly connected with the percent of compounds adsorbed onto modified magnetic nanoparticles