Electrocardiograph　signals　reflect　the　current　state　of　the　heart　and　have　great　significance　to　the　clinical　diagnosis　of　the　heart.　Convolutional　neural　networks　perform　excellently　in　electrocardiograph　pattern　recognition.　However,　CNNs　processing　ECG　signals　need　to　convert　them　from　1D　to　2D,　leading　to　additional　circuit　and　time　costs　in　hardware.　Here,　a　convolution　organic　transistor　(COT)　is　proposed　for　monitoring　the　ECG　with　CNNs.　Based　on　the　surface　electric　field　effect　and　trap　effect,　COT　can　directly　process　1D　ECG　data　without　complex　preprocessing.　It　can　complete　the　convolution　calculation　of　ECG　signals　≈20 000　times　per　second　in　theory　and　reduce　the　number　of　devices　by　83%　compared　to　conventional　arrays.　Further,　actual　ECG　signals　are　measured　and　input　into　the　COT,　which　can　initially　recognize　the　type　of　ECG　abnormality.　Finally,　a　point　calculation　detection　system　is　established　with　96.2%　recognition　accuracy　in　the　five‐heartbeat　classification　task　by　combining　the　1D　CNN.