We present an initial exploration of a fully cost-driven design. A design criterion was proposed that represented the minimum expected cost of an early phase clinical study, where costs include resource use as well as study failure. The design was based on attainment of a target concentration in a cohort of study participants. The model and parameter values arose from a previous population pharmacokinetic analysis of a phase I study. The resulting design naturally balanced the cost and the success rate of an early phase clinical study, without the need to define arbitrary constraints on the design space.
Activity recognition in smart homes aims to infer the particular activities of the inhabitant, the aim being to monitor their activities and identify any abnormalities, especially for those living alone. In order for a smart home to support its inhabitant, the recognition system needs to learn from observations acquired through sensors. One question that often arises is which sensors are useful and how many sensors are required to accurately recognise the inhabitant's activities? Many wrapper methods have been proposed and remain one of the popular evaluators for sensor selection due to its superior accuracy performance. However, they are prohibitively slow during the evaluation process and may run into the risk of overfitting due to the extent of the search. Motivated by this characteristic, this paper attempts to reduce the cost of the evaluation process and overfitting through tree alignment. The performance of our method is evaluated on two public datasets obtained in two distinct smart home environments.
Osteoporosis is a serious bone disease that affects many people worldwide. Various drugs have been used to treat osteoporosis. However, these drugs may cause severe adverse events in patients. Adverse drug events are harmful reactions caused by drug usage and remain one of the leading causes of death in many countries. Predicting serious adverse drug reactions in the early stages can help save patients' lives and reduce healthcare costs. Classification methods are commonly used to predict the severity of adverse events. These methods usually assume independence among attributes, which may not be practical in real-world applications. In this paper, a new attribute weighted logistic regression is proposed to predict the severity of adverse drug events. Our method relaxes the assumption of independence among the attributes. An evaluation was performed on osteoporosis data obtained from the United States Food and Drug Administration databases. The results showed that our method achieved a higher recognition performance and outperformed baseline methods in predicting the severity of adverse drug events.
Sensor-based human activity recognition has been extensively studied. Systems learn from a set of training samples to classify actions into a pre-defined set of ground truth activities. However, human behaviours vary over time, and so a recognition system should ideally be able to continuously learn and adapt, while retaining the knowledge of previously learned activities, and without failing to highlight novel, and therefore potentially risky, behaviours. In this paper, we propose a method based on compression that can incrementally learn new behaviours, while retaining prior knowledge. Evaluation was conducted on three publicly available smart home datasets.
This is a review of the first 100 coronary stents implanted for the indications of PTCA restenosis, acute or threatened closure and De Novo lesions. The success rates were high and complications rates were low. Subacute stent thrombosis rates were low and stenting for De Novo lesions in > 3.0 mm arteries provide the best short and long term results. Six months restenosis rate was low. Stent assisted high pressure balloon angioplasty is an important advance.