In Taiwan, thousands of people are sent to intensive care unit (ICU) for all kinds of reasons every day. Because there are many kinds of bacteria, a ventilated patient is easily infected, causing pneumonia, septicemia shock, cardiopulmonary failure, and even death. According to the doctor's experience, the standard operating procedure (SOP) is to take a chest X-ray, draw blood, perform a sputum culture, and then treat the patient with antibiotics of the speculated microorganisms. No one knows if the given antibiotics are suitable because specimen culture needs five days, which may be the most crucial time for the patient to survive. Since the microorganisms generate gases during metabolism, these gases could be excellent candidates for early prediction and real-time detection of pneumonia. If we can install a chip that can inhale to the respirator, there is a huge chance to achieve real-time detection for every ventilated patient in the ICU and reduce the death rate due to this problem. This project will proceed in clinical inspection and engineering methods and integrate them to test the proposed system in clinical trials. In clinical inspection, patients' exhaled gases will be collected and analyzed; at the same time, sputum culture will be performed to assure the microorganisms. This will establish a database of the relationship between the exhaled gases and the microorganisms. In engineering, we will proceed with researches on sensors, signal processing, algorithm, and system integration, to design, develop, and integrate a low-power, low-voltage electronic nose system-on-chip (SoC). The electronic nose SoC will be integrated into the expired circuit of the respirator for clinical trials to verify the feasibility of early prediction and real-time detection of the microorganisms of pneumonia. Objectives: Early prediction and real-time detection of the microorganisms of pneumonia in ventilated-patients by an electronic nose SoC. Solution: Develop, and integrate a low-voltage low-power electronic nose system-on-chip (SoC) comprising sensors, signal processing, and algorithms. Technical benchmark: A low-voltage low-power electronic nose system-on-chip (SoC) with recognizable detection.