Epilepsy is a major neurological disorder causing sudden or recurrent seizures. The seizures cause an abrupt loss of motor control and prove to impact fatal consequences or hazardous situations in patients. These seizures occur without any warning, forcing the patients to develop an anxiety in their daily life. This proposed system shall predict and alarm the patients or care-takers to improve the patients’ quality of life and safety. An EEG (Electroencephalogram) electrode cap is attached to the patient’s scalp connected to a battery powered flat, shockproof, battery powered system. The system is attached to patients back as parachute is attached in air divers. The system uses grid SVM (Support Vector Machine) technique to analyze the EEG data acquired for early prediction of seizures. The patient is subjected to continuous analysis for a week or minimum of three seizure occurrences in the clinic. The inter-ictal pattern just before the pre-ictal period or pre-ictal pattern is stored in the pattern storage module a reference pattern. Now the adaptive system is ready to use out of clinic or hospital care units. The periodically generated pattern shall be checked with reference patterns. Alarm is given when pattern matching is found and band power of beta waves is high. This system is adaptive due to pattern recognition technique.