Aims: Heart rate analysis can be used for automatic seizure detection in a home environment. The downside of heart rate based seizure detection algorithms is their strong dependency on the patient’s heart rate characteristics. This leads to mediocre performance if a patient-independent algorithm is used. Fully patient-specific algorithms are rarely used in practice due to the low amount of available patient-specific (seizure) data. The aim is to make a personalized seizure detection algorithm with a limited amount of patient-specific data using supervised transfer learning. Methods: Transfer learning is a machine learning technique that allows to train classifiers from a limited amount of data if a reference classifier is available. In this study, a state-of-the-art patient-independent support vector machine classifier is used as a reference. The patient-specific classifiers are then obtained using this supervised transfer learning approach, for which 1-2 days of annotated ECG data are used for training. The method is evaluated on 6 temporal lobe epilepsy patients during 207 hours of day and night data, containing in total 74 complex partial and secondary generalized seizures. Results: The patient-independent algorithm resulted in on average 93.9% sensitivity and 2.5 false alarms per hour. After applying the personalized algorithms obtained through transfer learning, the false alarm rate dropped to 1.3 false alarms per hour, but also the sensitivity dropped slightly to 89.8%. The results show that the algorithm can already adapt to patient-specific heart rate characteristics by using only a couple of days of annotated data, leading to almost 50% less false alarms compared to the patient-independent classifier without significantly reducing the sensitivity. Conclusion: Transfer learning allows to robustly personalize heart rate based seizure detection with a limited amount of patient-specific data. It allows to strongly improve the performance fast, increasing the usability in practice for a real-time seizure warning system.