Add BrightSurf on Google Email

Computers using linguistic clues to deduce photo content

Researchers at Disney Research and UC Davis have developed a method for computer vision programs to understand spatial relationships in images based on caption sentence structure. This approach enables accurate visual localizations for language inputs, outperforming baseline systems that do not consider natural language structure.

A computer that reads body language

Researchers at Carnegie Mellon University developed a computer that can understand body poses and movements of multiple people in real-time, including finger poses, using a single camera and laptop. This enables new ways for people and machines to interact, such as more natural communication with computers by pointing at objects.

Hospital, office physicians have differing laments about electronic records

A new study reveals widespread physician dissatisfaction with electronic health records (EHRs), with hospital-based physicians expressing concerns about reduced patient contact time and office-based physicians worrying about the quality of their interactions. Despite these challenges, some physicians have found ways to minimize disrupt...

SourceBrown University·JournalJournal of Innovation in Health Informatics·DateJul 5, 2017

World's thinnest hologram paves path to new 3-D world

The RMIT team has developed a nano-hologram that is simple to make, can be seen without 3D goggles and is 1000 times thinner than a human hair. The discovery could transform industries such as medical diagnostics, education, data storage, defence and cyber security with the potential to display a wealth of data.

SourceRMIT University·JournalNature Communications·DateMay 18, 2017

Hospitals must be prepared for ransomware attacks

Ransomware attacks on hospitals have increased fourfold from 2015 to 2016, with the amount of money paid to hackers rising to $1bn. Hospitals can take steps to prevent these attacks by implementing digital hygiene measures and frequent backups to protect their data.

SourceBMJ Group·JournalThe BMJ·DateMay 10, 2017

Hand that sees offers new hope to amputees

A new generation of prosthetic limbs with built-in cameras can now automatically pick up objects, eliminating the need for manual control. Researchers at Newcastle University have developed a 'hand that sees' using computer vision and neural networks, which can assess object shape and size in real-time to trigger precise movements.

SourceNewcastle University·JournalJournal of Neural Engineering·DateMay 3, 2017

Gaming helps personalized therapy level up

Researchers at Penn State have developed a method to create personalized mental and physical therapy regimens using gaming features. Gamified applications with scoring systems, avatars, and in-game rewards led to significantly fewer mistakes and higher performance than non-gamified applications.

SourcePenn State·JournalComputers in Human Behavior·DateApr 19, 2017

Computer trained to predict which AML patients will go into remission, which will relapse

Researchers developed a computer machine-learning model that accurately predicts which AML patients will go into remission following treatment. The model was trained using bone marrow data and medical histories of AML patients, achieving 100% accurate predictions for remission and 90% accurate predictions for relapse.

SourceIndiana University-Purdue University Indianapolis School of Science·JournalIEEE Transactions on Biomedical Engineering·DateFeb 9, 2017

Computer work dominates physician workday

A recent study by the American College of Physicians found that physicians spend significantly more time on computer activities than direct patient interaction, taking up about half their workday. This has led to concerns over physician satisfaction, patient education, and increased malpractice risks.

SourceAmerican College of Physicians·JournalAnnals of Internal Medicine·DateJan 30, 2017

Engineers eat away at Ms. Pac-Man score with artificial player

Researchers at Cornell University developed an artificial Ms. Pac-Man player that achieved a laboratory score of 43,720, surpassing the existing high score for computerized play. The player uses a decision-tree approach and demonstrates accuracy in predicting ghost movements with 94.6-percent accuracy.

SourceCornell University·JournalIEEE Transactions on Computational Intelligence and AI in Games·DateJan 23, 2017