Kaohsiung Medical University | Laser Aim for Internal Fracture Fixation

In general, 3D photo-curing printed bio-ceramic application could not meet the expectations of mechanical properties of printed objects. By admixing a combustible reverse negative thermoresponsive hydrogel (poly(N-isopropylacrylamide)-based), this problem can be solved. Sintering densification is expected via free volume contraction, which will increase the mechanical properties after the formation of the porous bio-ceramics. These bio-ceramic devices could offer additional benefits for bone or prosthodontics. This photo-curing technology seeks to facilitate the fabrication of more precise and complex shapes using 3D printing of ceramic know-how.

First of all, the image classification technique and retinal segmentation were put together to generate an automatic classification for CKD patient. It saves time on filtering unqualified image. Moreover, with Resnet50 image classification, it improves the efficiency of image processing. After preprocessing step, AI system is used analyzing retinal image and classifying CKD stage, which is aim to predict high risk group of DKD in patients with diabetes.

A new technology is developed to assist in diagnosing ADHD through a wearable device. This study found that the use of a smartwatch to detect the variance and zero-crossing rate of movements in children with ADHD is a very promising method to assist in the diagnosis. The smartwatch can collect activity data during the assessment period, which includes the three-axis angular acceleration and three-axisacceleration of the subject's movements, to obtain six sets of momentum data. At least one momentum data will be used for detection, and it is analyzed whether its value exceeds the value of standard data. The smartwatch can be used to solve the problems of the traditional diagnosis of ADHD based on the behavior rating scales, which is over-subjective, and inaccurate.

According to the guidelines of the European Association of Urology, 15% of cases of nephrolithiasis are caused by uric acid stones, which can be dissolved with medication and do not require surgery. However, currently, surgery is the only way to analyze and determine whether a stone is a uric acid stone. In the past, we found that there are eight simple clinical factors highly correlated with uric acid stones. Therefore, we have combined artificial intelligence with these eight clinical factors to accurately predict whether a patient's kidney stone is a uric acid stone after being diagnosed with nephrolithiasis.

Blepharoptosis refers to the weakening or loss of the Levator muscle of the upper eyelids or Muller's muscle function, resulting in the pupil (iris) being covered by the eyelid, thereby affecting the visual field and appearance. The conventional ptosis assessment is considered a subjective, unreliable, and invalid method as a manual measurement and evaluation of various parameters are performed by doctors using a measuring ruler. Hence, it is difficult to judge and evaluate the benefits of diagnosis and treatment before and after surgery. The Automatic Blepharoptosis Detection System (ABDS) is an automated image measurement equipment featuring clinical objectivity, accuracy, and consistency. It can accurately detect and record all kinds of blepharoptosis parameters, as listed below, that helps doctors to assess the severity of blepharoptosis, and further determine the treatment method. ABDS provides objective information when applying for health insurance benefits. It can also be used as a convenient tool for face-to-face communication with patients and their family members. The most important is that ABDS establish a standard procedure for diagnosis and treatment of blepharoptosis.

The association of Alzheimer's Disease International (ADI) estimates that there are more than 500 million people with dementia worldwide in 2019. By 2050, it is expected that the number of people with dementia will grow up to 100 million. Dementia-related costs currently stand at $1 trillion a year and are expected to double by 2030. AD accounts around 60 % of these demented patients, and it can not be cured currently. Given to heterogeneously therapeutic outcomes and no cured medicines for AD, there is no available makers to reflect clinical outcomes and provide adequate managements with referring to clinical big data. Different stage of AD has its needs. If we can predict the next stage for a AD patient, we can have appropriate managements to benefit patients and their caregivers. For above sakes, we have computed and integrated several clinical data pools to provide a predictive tool for AD outcomes.

New and improved drugs have been developed for treating different stages of prostate cancer. These drugs can extend the time people with prostate cancer live, but over time, some patients develop resistance to them. The goal of this research is to identify gene signatures for good and poor prognosis through customized Affymetrix transcriptome chips in a fast and highly reproducible manner. Prostate cancer patients from different medical centers were studied. A customized Affymetrix transcriptome chip was used in the analysis of blood samples before, during, and after drug treatment. By examining the patterns of gene expression and using various models, key genes related to drug response were identified. With this information, an online tool was created to recommend the most effective drugs for individual patients based on their genetic profiles.

Although there are related patents using body fluids to detect metabolites for predicting the disease occurrence, those disclosed metabolites are different from this technology, and this patent focuses on breast cancer-related metabolites to predict the recurrence and prognosis of breast cancer patients. The concentration of metabolites can be further compared with a threshold value and arranged to achieve the best metabolite panel with the best sensitivity, specificity and AUC (Area Under ROC Curve) under Receiver Operator Characteristic curve (ROC curve).

Cardiovascular disease is a major global cause of death. Blood pressure monitoring is crucial, but current methods have limitations. We propose a non-invasive, accurate, and continuous method for predicting arterial pressure waveforms using composite algorithm. This enhances accuracy and real-time prediction, allowing medical professionals to monitor patients and provide timely treatment. The non-invasive approach improves patient comfort and avoids risks. Our deep learning model reduces costs, human error, and complies with industry standards. Our invention will revolutionize cardiovascular disease management by using TCN-GRU algorithms, leading to precise and timely intervention. This invention could become a standard tool for cardiovascular function monitoring, greatly impacting healthcare.