Using data mining for prediction of hospital length of stay: an application of the CRISP-DM Methodology

Hospitals are nowadays collecting vast amounts of data related with patient records. All this data hold valuable knowledge that can be used to improve hospital decision making. Data mining techniques aim precisely at the extraction of useful knowledge from raw data. This work describes an implementation of a medical data mining project approach based on the CRISP-DM methodology. Recent real-world data, from 2000 to 2013, were collected from a Portuguese hospital and related with inpatient hospitalization. The goal was to predict generic hospital Length Of Stay based on indicators that are commonly available at the hospitalization process (e.g., gender, age, episode type, medical specialty). At the data preparation stage, the data were cleaned and variables were selected and transformed, leading to 14 inputs. Next, at the modeling stage, a regression approach was adopted, where six learning methods were compared: Average Prediction, Multiple Regression, Decision Tree, Artificial Neural Network ensemble, Support Vector Machine and Random Forest. The best learning model was obtained by the Random Forest method, which presents a high quality coefficient of determination value (0.81). This model was then opened by using a sensitivity analysis procedure that revealed three influential input attributes: the hospital episode type, the physical service where the patient is hospitalized and the associated medical specialty. Such extracted knowledge confirmed that the obtained predictive model is credible and with potential value for supporting decisions of hospital managers.

Child abuse monitor system model: a health care critical knowledge monitor system

The Childhood protection is a subject with high value for the society, but, the Child Abuse cases are difficult to identify. The process from suspicious to accusation is very difficult to achieve. It must configure very strong evidences. Typically, Health Care services deal with these cases from the beginning where there are evidences based on the diagnosis, but they aren’t enough to promote the accusation. Besides that, this subject it’s highly sensitive because there are legal aspects to deal with such as: the patient privacy, paternity issues, medical confidentiality, among others. We propose a Child Abuses critical knowledge monitor system model that addresses this problem. This decision support system is implemented with a multiple scientific domains: to capture of tokens from clinical documents from multiple sources; a topic model approach to identify the topics of the documents; knowledge management through the use of ontologies to support the critical knowledge sensibility concepts and relations such as: symptoms, behaviors, among other evidences in order to match with the topics inferred from the clinical documents and then alert and log when clinical evidences are present. Based on these alerts clinical personnel could analyze the situation and take the appropriate procedures.

Improving quality of medical service with mobile health software

An increasing number of m-Health applications are being developed benefiting health service delivery. In this paper, a new methodology based on the principle of calm computing applied to diagnostic and therapeutic procedure reporting is proposed. A mobile application was designed for the physicians of one of the Portuguese major hospitals, which takes advantage of a multi-agent interoperability platform, the Agency for the Integration, Diffusion and Archive (AIDA). This application allows the visualization of inpatients and outpatients medical reports in a quicker and safer manner, in addition to offer a remote access to information. This project shows the advantages in the use of mobile software in a medical environment but the first step is always to build or use an interoperability platform, flexible, adaptable and pervasive. The platform offers a comprehensive set of services that restricts the development of mobile software almost exclusively to the mobile user interface design. The technology was tested and assessed in a real context by intensivists.

Development of polymer wicks for the fabrication of bio-medical sensors

Polymer based wicking structures were fabricated by sintering powders of polycarbonate (PC), ultra-high molecular weight polyethylene and polyamide 12, aiming at selecting a suitable material for an innovative electroencephalography (EEG) bio-electrode. Preliminary experiments showed that PC based wicks displayed the best mechanical properties, therefore more detailed studies were carried out with PC to evaluate the influence of powder granulometry and processing parameters (pressure, temperature and time) on the mechanical properties, porosity, mean pore radius and permeability of the wicks. It was concluded that the mechanical properties are significantly enhanced by increasing the processing time and pressure, although at the expense of a significant decrease of porosity and mean pore diameter (and thus permeability), particularly for the highest applied pressures (74kPa). However, a good compromise between porosity/permeability and mechanical properties could be obtained by sintering PC powders of particle sizes below 500μm at 165°C for 5min, upon an applied pressure of 56kPa. Moreover, PC proved to be chemically stable in contact with an EEG common used disinfectant. Thus, wicking structures with appropriate properties for the fabrication of reusable bio-electrodes could be fabricated from the sintering of PC powders.

DOTA-based Ga(III) and Gd(III) chelates for medical imaging (PET, SPECT and MRI)

PhD Thesis in Sciences Specialization in Chemistry

Influence of saliva and mucin on the adhesion of Candida oral clinical isolates

Objectives: This research work intends to clarify the role of artificial saliva, in particularly the role of mucin, a salivary protein, on the surface properties and adhesion ability of Candida spp. oral clinical isolates to abiotic surfaces. Methods: Four oral clinical isolates of Candida spp. were used: two Candida albicans strains (AC; AM) and two Candida parapsilosis strains (AD; AM2). The strains were isolated from patients using oral prosthesis. The microorganisms were cultured in the absence or presence of mucin and artificial saliva, and their adhesion to an abiotic surface (coated with mucin and artificial saliva) was evaluated. Results: The presence of mucin per se onto the abiotic surface decreased the adhesion of all strains, although the combination of mucin with artificial saliva had reduced this effect. No direct correlation between adhesion and the surface free energies of adhesion of the microorganisms was found. Significance: Candida spp. were human commensal microorganisms that became pathogenic when the host immune defenses were compromised. Medical devices were colonized by Candida spp. particularly, oral prostheses, which might lead to the degradation of the prostheses and systemic infections. The salivary secretions that constantly cover the oral cavity influenced Candida spp. adhesion process. Therefore, it was important to understand the interactions between Candida spp., salivary proteins and the characteristic of oral prosthesis when developing materials for oral prostheses.

Level of compliance in contact lens wearing medical doctors in Nepal

PURPOSES: To determine the level of compliance and major non-compliant behaviors in contact lens (CL) wearing medical doctors (MDs) and to compare it with age matched CL wearing normal subjects with no medical background (NS). METHODS: Thirty-nine current CL wearing MDs, who were prescribed CLs in Nepal Eye Hospital, Kathmandu, Nepal, between 2007 and 2011, were interviewed on ten modifiable compliant behaviors regarding lens care and maintenance. The level of compliance and the rate of non-compliance for each behavior were determined and compared with NS. RESULTS: Level of compliance was good, average and poor in 35.9%, 48.7% and 15.4% of MDs, respectively. There was no significant difference in compliance between MDs and NS (p=0.209). Level of compliance was not associated with age, gender and duration of lens wear (p>0.05). Compliance rate varied according to different behaviors, achieving a good compliance level of 95% for hand hygiene, avoidance of water contact and not sleeping with lenses. There was poor compliance for topping up solution (53.8%) and lens case replacement (15.4%). CONCLUSION: About one third of MDs had a good level of compliance. Level of compliance and compliance rate of different behaviors were similar in MDs and NS. Periodic lens case replacement was the most neglected behavior in CL wearers for this region.

Development of procedures for the design, optimization and manufacturing of customized orthopaedic and trauma implants: Geometrical/anatomical modelling from 3D medical imaging

Tese de Doutoramento (Programa Doutoral em Engenharia Biomédica)

Incorporation of antimicrobial peptides on functionalized cotton gauzes for medical applications

A large group of low molecular weight natural compounds that exhibit antimicrobial activity has been isolated from animals and plants during the past two decades. Among them, peptides are the most widespread resulting in a new generation of antimicrobial agents with higher specific activity. In the present study we have developed a new strategy to obtain antimicrobial wound-dressings based on the incorporation of antimicrobial peptides into polyelectrolyte multilayer films built by the alternate deposition of polycation (chitosan) and polyanion (alginic acid sodium salt) over cotton gauzes. Energy dispersive X ray microanalysis technique was used to determine if antimicrobial peptides penetrated within the films. FTIR analysis was performed to assess the chemical linkages, and antimicrobial assays were performed with two strains: Staphylococcus aureus (Gram-positive bacterium) and Klebsiella pneumonia (Gram-negative bacterium). Results showed that all antimicrobial peptides used in this work have provided a higher antimicrobial effect (in the range of 4 log–6 log reduction) for both microorganisms, in comparison with the controls, and are non-cytotoxic to normal human dermal fibroblasts at the concentrations tested.

Predicting triage waiting time in maternity emergency care by means of data mining

Healthcare organizations often benefit from information technologies as well as embedded decision support systems, which improve the quality of services and help preventing complications and adverse events. In Centro Materno Infantil do Norte (CMIN), the maternal and perinatal care unit of Centro Hospitalar of Oporto (CHP), an intelligent pre-triage system is implemented, aiming to prioritize patients in need of gynaecology and obstetrics care in two classes: urgent and consultation. The system is designed to evade emergency problems such as incorrect triage outcomes and extensive triage waiting times. The current study intends to improve the triage system, and therefore, optimize the patient workflow through the emergency room, by predicting the triage waiting time comprised between the patient triage and their medical admission. For this purpose, data mining (DM) techniques are induced in selected information provided by the information technologies implemented in CMIN. The DM models achieved accuracy values of approximately 94% with a five range target distribution, which not only allow obtaining confident prediction models, but also identify the variables that stand as direct inducers to the triage waiting times.

Optimization techniques to detect early ventilation extubation in intensive care units

The decision support models in intensive care units are developed to support medical staff in their decision making process. However, the optimization of these models is particularly difficult to apply due to dynamic, complex and multidisciplinary nature. Thus, there is a constant research and development of new algorithms capable of extracting knowledge from large volumes of data, in order to obtain better predictive results than the current algorithms. To test the optimization techniques a case study with real data provided by INTCare project was explored. This data is concerning to extubation cases. In this dataset, several models like Evolutionary Fuzzy Rule Learning, Lazy Learning, Decision Trees and many others were analysed in order to detect early extubation. The hydrids Decision Trees Genetic Algorithm, Supervised Classifier System and KNNAdaptive obtained the most accurate rate 93.2%, 93.1%, 92.97% respectively, thus showing their feasibility to work in a real environment.