The Machine Learning field has picked up its push in any space of examination and just as of late has turned into a dependable apparatus in the therapeutic area. The experiential area of programmed learning is utilized as a part of assignments, for example, restorative choice bolster, medicinal imaging, protein-protein collaboration, extraction of therapeutic information, and for general patient administration care. Machine Learning (ML) is imagined as a device by which PC based frameworks can be incorporated in the medicinal services field to show signs of improvement, all around sorted out therapeutic consideration. It depicts a ML-based strategy for building an application that is fit for distinguishing and dispersing social insurance data. It concentrates sentences from distributed medicinal papers that say maladies and medications, and distinguishes semantic relations that exist in the middle of ailments and medicines. This paper proposes a new way of information retrieval. A clustered similarity approach is used to overcome the previous approach’s drawbacks. Results are obtained and the proposed work is much better than existing techniques.