Machine learning techniques to examine large patient databases

Meyfroidt, Geert; Güiza, Fabián; Ramon, Jan; Bruynooghe, Maurice

doi:10.1016/j.bpa.2008.09.003

Cited by 89 publications

(69 citation statements)

References 40 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…However, machine-learning techniques have been used extensively in medicine, 21 in gene expression studies, [22][23][24] for classification of cardiac arrhythmias, 25 for predicting morbidity after coronary artery bypass surgery, 26 and for predicting when weaning from ventilator support should begin. 27 Gaussian processes have been applied in adults with ALI to model the pressure-volume curve to titrate PEEP.…”

Section: Discussionmentioning

confidence: 99%

Algorithms to Estimate P_aCO₂and pH Using Noninvasive Parameters for Children with Hypoxemic Respiratory Failure

et al. 2013

View full text Add to dashboard Cite

BACKGROUND: Ventilator management for children with hypoxemic respiratory failure may benefit from ventilator protocols, which rely on blood gases. Accurate noninvasive estimates for pH or P aCO 2 could allow frequent ventilator changes to optimize lung-protective ventilation strategies. If these models are highly accurate, they can facilitate the development of closed-loop ventilator systems. We sought to develop and test algorithms for estimating pH and P aCO 2 from measures of ventilator support, pulse oximetry, and end-tidal carbon dioxide pressure (P ETCO 2 ). We also sought to determine whether surrogates for changes in dead space can improve prediction. METHODS: Algorithms were developed and tested using 2 data sets from previously published investigations. A baseline model estimated pH and P aCO 2 from P ETCO 2 using the previously observed relationship between P ETCO 2 and P aCO 2 or pH (using the Henderson-Hasselbalch equation). We developed a multivariate gaussian process (MGP) model incorporating other available noninvasive measurements. RESULTS: The training data set had 2,386 observations from 274 children, and the testing data set had 658 observations from 83 children. The baseline model predicted P aCO 2 within ؎ 7 mm Hg of the observed P aCO 2 80% of the time. The MGP model improved this to ؎ 6 mm Hg. When the MGP model predicted P aCO 2 between 35 and 60 mm Hg, the 80% prediction interval narrowed to ؎ 5 mm Hg. The baseline model predicted pH within ؎ 0.07 of the observed pH 80% of the time. The MGP model improved this to ؎ 0.05. CONCLUSIONS: We have demonstrated a conceptual first step for predictive models that estimate pH and P aCO 2 to facilitate clinical decision making for children with lung injury. These models may have some applicability when incorporated in ventilator protocols to encourage practitioners to maintain permissive hypercapnia when using high ventilator support. Refinement with additional data may improve model accuracy.

show abstract

Section: Discussionmentioning

confidence: 99%

Algorithms to Estimate P_aCO₂and pH Using Noninvasive Parameters for Children with Hypoxemic Respiratory Failure

et al. 2013

View full text Add to dashboard Cite

show abstract

“…The major requirement is to devise systems that detect patient-specific physiological states in real-time using minimally invasive and low power devices. The development in machine-learning algorithms that are capable of exploiting statistical properties in the data to model specific correlations will facilitate more accurate decision making by healthcare experts or devices [26]. For disease monitoring using machine learning models, prior pre-processing is as important, or even more important than the use of the machine learning models themselves [27].…”

Section: B Supporting Technologymentioning

confidence: 99%

Complementary Detection for Hardware Efficient On-Site Monitoring of Parkinsonian Progress

Mohammed

Demosthenous

2018

IEEE J. Emerg. Sel. Topics Circuits Syst.

View full text Add to dashboard Cite

Abstract-The progress of Parkinson's disease (PD) in patients is conventionally monitored through follow-up visits. These may be insufficient for clinicians to obtain a good understanding of the occurrence and severity of symptoms in order to adjust therapy to the patients' needs. Portable platforms for PD diagnostics can provide in-depth information, thus reducing the frequency of face-to-face visits. This paper describes the first known on-site PD detection and monitoring processor. This is achieved by employing complementary detection which uses a combination of weak k-NN classifiers to produce a classifier with a higher consistency and confidence level than I. INTRODUCTIONARKINSON'S disease (PD) has complex mechanisms [1], and to optimize therapy, a better understanding of its dynamics is required. Currently, the standard for diagnosing and monitoring parkinsonian progress in patients is observation of visual feedbacks from them [2]. These may be insufficient since it is only monitored during follow-up visits. Research in disease monitoring has ranged from using mobile devices that have short message service, web-based applications and Bluetooth capability to measure the frequency of symptom onset so that medical interventions could be delivered or better diagnosis can be made [3]. These systems can be implemented on software applications running on the patient's commercial smartphone and connected to the clinician's information systems [4], for example, the WebBioBank, a web-based system for collecting clinical and neurophysiological data [5]. It is specifically created for deep brain stimulation (DBS) management, and can also be connected to the patient's mobile applications so that it can safely be used for web-based tele-monitoring and caregiver support [3]. Such disease monitoring can be used to provide a more refined therapy and for biomarker selection based on patient data collected.The power required to transmit data in neural signal processing systems dominates that for recording and data conversion [6] and offline processing based on transmitting raw time series data as suggested in [3], [5], is an inefficient approach. Based on the power and bandwidth constraints involved in continuously sending neural signals, it will be more resource efficient to periodically send patient progress as state estimates after on-site and online analysis. Such an integrated platform for on-site and online analysis and monitoring of PD signals is still unavailable. For on-site and online analysis, there is a need to develop miniaturized realtime platforms that could monitor disease progress. These specialized hardware platforms would facilitate mobile diagnostics for better disease management. Portable platforms for PD diagnostics could provide more in-depth information and reduce the number of face-to-face visits required to optimize therapy. In PD monitoring, the aim is to provide long-term monitoring of the patient's condition for clinicians to better understand the symptoms so that therapy could be more accuratel...

show abstract

“…Lynch et al [29] predicted lung cancer patient survival via supervised machine learning algorithms. Machine learning was also used in eHealth for analyzing patient's health data, predicting diseases, enhancing the productivity of technology or devices used for service providing, and so on [30], [31]. However, a very few studies [32]- [34] was found to predict the usage of eHealth among its target patients.…”

Section: Background and Objectivementioning

confidence: 99%

Predicting rural patients� use of eHealth through supervised machine learning algorithms: A study on Portable Health Clinic in Bangladesh (Preprint)

Hossain¹,

Yokota²,

Fukuda³

et al. 2018

Preprint

View full text Add to dashboard Cite

Background: Predictive analytics through machine learning has been extensively using across industries including eHealth and mHealth for analyzing patient's health data, predicting diseases, enhancing the productivity of technology or devices used for providing healthcare services and so on. However, not enough studies were conducted to predict the usage of eHealth by rural patients in developing countries.

show abstract

Machine learning techniques to examine large patient databases

Cited by 89 publications

References 40 publications

Algorithms to Estimate P_aCO₂and pH Using Noninvasive Parameters for Children with Hypoxemic Respiratory Failure

Algorithms to Estimate P_aCO₂and pH Using Noninvasive Parameters for Children with Hypoxemic Respiratory Failure

Complementary Detection for Hardware Efficient On-Site Monitoring of Parkinsonian Progress

Predicting rural patients� use of eHealth through supervised machine learning algorithms: A study on Portable Health Clinic in Bangladesh (Preprint)

Contact Info

Product

Resources

About

Machine learning techniques to examine large patient databases

Cited by 89 publications

References 40 publications

Algorithms to Estimate PaCO2and pH Using Noninvasive Parameters for Children with Hypoxemic Respiratory Failure

Algorithms to Estimate PaCO2and pH Using Noninvasive Parameters for Children with Hypoxemic Respiratory Failure

Complementary Detection for Hardware Efficient On-Site Monitoring of Parkinsonian Progress

Predicting rural patients� use of eHealth through supervised machine learning algorithms: A study on Portable Health Clinic in Bangladesh (Preprint)

Contact Info

Product

Resources

About

Algorithms to Estimate P_aCO₂and pH Using Noninvasive Parameters for Children with Hypoxemic Respiratory Failure

Algorithms to Estimate P_aCO₂and pH Using Noninvasive Parameters for Children with Hypoxemic Respiratory Failure