In a study published in Nature Machine Intelligence, researchers at Ohio State University show how artificial intelligence (AI) can follow clinical trials to identify drugs for repurposing, a solution that can help advance innovative treatments.

Repurposing drugs is legal and not unusual. When doctors prescribe drugs that have been approved by the Food and Drug Administration (FDA) for purposes different from what is printed on the labels, the drugs are being used "off - label". Just because a drug is FDA - approved for a specific type of disease does not prevent it from having possible benefits for other purposes.

For example, Metformin, a drug that is FDA - approved for treating type 2 diabetes, is also used to treat PCOS (a disease of women), and other diseases. Trazodone, an anti - depressant with FDA - approval to treat depression, is also prescribed by doctors to help treat patients with sleep issues.

The Ohio State University research team created an AI deep learning model for predicting treatment probability with patient data including the treatment, outcomes , and potential confounders (干扰因素).

Confounders are related to the exposure and outcome. For example , a connection is identified between music festivals and increases in skin rashes (红疹). Music festivals do not directly cause skin rashes. In this case, one possible confounding factor between the two may be outdoor heat, as music festivals tend to run outdoors when the temperature is high, and heat is a known cause for rashes. When working with real - world data, confounders could number in the thousands. AI deep learning is well-suited to find patterns in the complexity of potentially thousands of confounders.

The researcher team used confounders including population data and co-prescribed drugs. With this proof -of-concept, now clinicians have a powerful AI tool to rapidly discover new treatments by repurposing existing medications.

Anyone keen to view from on high the prosperous cityscape of Guangdong, may soon can do so from the cabin of a flying taxi. On October 13th the Civil Aviation (航空) Administration of China (CAAC) awarded a "type certificate" to the world's first electric vertical take-off and landing (eVTOL)taxi. And the small two-seater, called the EH216-S, was also permitted to fly without a pilot on board.

The EH216-s is made by EHang, a company based in Guangdong. It looks like a large consumer drone with a passenger bubble installed on top. Propulsion (推动力) is provided by 16 small rotors (转子), installed on the tips of eight arms that fold away when the vehicle is not in use, allowing it to park in small spaces.

The CAAC gave its approval after EHang had conducted more than 40,000 test flights, including with volunteer passengers in 18 cities across China. It also monitored the EH216-s to structural analysis and crash tests, and checked its ability to keep flying if one of its rotors fails. Regulators (监管机构) also inspected the wireless network EHang uses to link its flying taxis to a control centre on the ground. That allows backup pilots to land an aircraft by remote control if there is a problem.

EHang says its pilotless eVTOLs will be quieter than helicopters and much cheaper to operate, thanks to the ability to swap out an expensive pilot for a second paying passenger. The firm also thinks that doing away with pilots will make things safer. A computer's attention never swings, and its reaction operate at a high speed. And it is much easier to automate than driving,for there are fewer barriers and unexpected situations to navigate.

EHang has ambitions beyond China, too. It has carried out demonstration flights in Amèrica and Europe with the aim of obtaining type certificates in both markets. But their regulators both have signalled that they will allow autonomous flights only after the vehicles have proved themselves safe in human hands.

Artificial intelligence helps teachers identify which of their students have learning difficulties, according to a new study.

Teachers make more accurate assessments of learning difficulties among their students if they are given Al-generated feedback(反馈) rather than an "expert solution" written by a qualified professional. Researchers believe that because the AI analyzed the teacher's own work, it's easier for the teacher to understand than a "model answer" prepared beforehand.

"Teachers play a critical role in recognizing the signs of disorders and learning difficulties in pupils and referring them to specialists," said Riikka Hofmann. "Unfortunately, many of them feel that they have not had sufficient opportunity to practice these skills. AI could provide an extra level of individualized feedback to help them develop these essential competences. "

Trainee teachers were asked to assess fictionalized students for potential learning difficulties, based on evidence including examples of their work, school behavior records and written texts of conversations with parents. Half of the trainees received the "expert solution", typical of the material given to trainee teachers, while the other half received AI-generated feedback on their approach, highlighting where they could improve. The trainees then completed similar follow-up assessments, and were graded both on the accuracy of their diagnoses(诊断) and on how well they had used the evidence. Trainees who received the AI feedback scored significantly higher than those who worked with the pre-written expert solutions.

While the researchers said this does not mean AI is preferable to one-to-one feedback from a skilled expert, this is not always available for trainee teachers. "We are not arguing that AI should replace teacher-educators: new teachers still need expert guidance on how to recognize learning difficulties in the first place," said Dr Michael Sailer. "It does seem, however, that Al-generated feedback helped these trainees to focus on what they really needed to learn. When personal feedback is not readily available, it could be an effective substitute(代替物). "