• By Jake Chung / Staff writer, with CNA

A new zero-contact speech recognition system using artificial intelligence (AI) technology would help medical staff wearing personal protective equipment use computers, Kaohsiung Chang Gung Memorial Hospital said on Wednesday.

The hospital developed the system in collaboration with Taiwan AI Labs after it had problems dealing with a surge in people seeking services at the hospital during a domestic COVID-19 outbreak that started in May, it said.

The hospital realized that staff wearing the protective equipment had problems when performing certain tasks, such as updating people’s medical records, the hospital said.

While usually updating the records immediately when attending to a patient, doctors now took notes afterwards, which increased the risk of error, it said.

Especially staff at screening stations, who are required to wear protective suits, masks and gloves, found it difficult to accurately type on a keyboard, it added.

After hearing of the problems, Taiwan AI Labs reached out to the hospital’s emergency medicine department director, Chiu Yi-min (邱義閔), suggesting a system using AI-based speech recognition technology, the hospital said.

The newly installed system would allow staff to make changes to medical files in real time and integrate data from different sources in the hospital’s records, it said.

Hospital superintendent Wang Chih-hsi (王植熙) said that the cooperation came at the right time, as the hospital had just started implementing Big Data and AI-based technologies to provide better medical services.

The hospital would continue to collaborate with Taiwan AI Labs to train the speech recognition AI, with the aim that the technology understands all relevant details of a dialogue between doctor and patient, Wang said.

Hopefully the system can help doctors make diagnoses and assess the severity of patients’ symptoms, he said.

Taiwan AI Labs said it has since May invested considerably in COVID-19-related technology, including a social distancing app and a diagnosis software.