A frequency reconfigurable flexible monopole antenna for wearable application is demonstrated in this article. Initially, the antenna consists of a tuning fork-shaped patch for obtaining the operating frequency at 5.2 GHz and after that a meander-shaped element has been attached to lengthen the proposed structure for performing another resonating frequency at 2.45 GHz for biotelemetry application. Frequency agility is achieved by varying the switching mechanism of the PIN diodes. The antenna has accomplished a compact size of 40 mm × 36 mm × 1 mm which has been employed on textile substrate. Here, we use some of the machine learning models of supervised learning to aid in the creation of the structure focused on the enhancement of bandwidth. The parameters have been optimized with R-squared validation 0.97–0.99 and achieved mean square error is nearly 0.03–0.5 whereas the given inputs are operating frequencies, length, width, height, and dielectric properties of substrate and reflection coefficient of the designed structure. Multiple ML algorithms like GPR, SVR, ANNs, and RF Regression are used to predict the parameters by splitting the dataset into training and testing with the ratio of 60 and 40 and comparative study of the results is mentioned here. This antenna exhibits 4.75 dB gain at 2.45 GHz and 9.18 dB gain at 5.2 GHz. The antenna is investigated by placing it on the various parts of human body and the obtained results provide the safety and suitability of the proposed antenna for use in biomedical application whereas the specific absorption rate is satisfied the limit that is set by the Federal Communication Commission.