Patient-Specific Modelling in Drug Design, Development and Selection Including its Role in Clinical Decision-Making

Abstract Background Predictive modeling with longitudinal electronic health record (EHR) data offers great promise for accelerating personalized medicine and better informs clinical decision-making. Recently, deep learning models have achieved state-of-the-art performance for many healthcare prediction tasks. However, deep models lack interpretability, which is integral to successful decision-making and can lead to better patient care. In this paper, we build upon the contextual decomposition (CD) method, an algorithm for producing importance scores from long short-term memory networks (LSTMs). We extend the method to bidirectional LSTMs (BiLSTMs) and use it in the context of predicting future clinical outcomes using patients’ EHR historical visits. Methods We use a real EHR dataset comprising 11071 patients, to evaluate and compare CD interpretations from LSTM and BiLSTM models. First, we train LSTM and BiLSTM models for the task of predicting which pre-school children with respiratory system-related complications will have asthma at school-age. After that, we conduct quantitative and qualitative analysis to evaluate the CD interpretations produced by the contextual decomposition of the trained models. In addition, we develop an interactive visualization to demonstrate the utility of CD scores in explaining predicted outcomes. Results Our experimental evaluation demonstrate that whenever a clear visit-level pattern exists, the models learn that pattern and the contextual decomposition can appropriately attribute the prediction to the correct pattern. In addition, the results confirm that the CD scores agree to a large extent with the importance scores generated using logistic regression coefficients. Our main insight was that rather than interpreting the attribution of individual visits to the predicted outcome, we could instead attribute a model’s prediction to a group of visits. Conclusion We presented a quantitative and qualitative evidence that CD interpretations can explain patient-specific predictions using CD attributions of individual visits or a group of visits.

Download Full-text

A machine learning framework for automated diagnosis and computer-assisted planning in plastic and reconstructive surgery

Scientific Reports ◽

10.1038/s41598-019-49506-1 ◽

2019 ◽

Vol 9 (1) ◽

Cited By ~ 2

Author(s):

Paul G. M. Knoops ◽

Athanasios Papaioannou ◽

Alessandro Borghi ◽

Richard W. F. Breakey ◽

Alexander T. Wilson ◽

...

Keyword(s):

Machine Learning ◽

Decision Making ◽

Reconstructive Surgery ◽

Clinical Decision Making ◽

Specific Treatment ◽

Operation Time ◽

Clinical Decision ◽

Patient Specific ◽

Computer Assisted ◽

Plastic And Reconstructive Surgery

Abstract Current computational tools for planning and simulation in plastic and reconstructive surgery lack sufficient precision and are time-consuming, thus resulting in limited adoption. Although computer-assisted surgical planning systems help to improve clinical outcomes, shorten operation time and reduce cost, they are often too complex and require extensive manual input, which ultimately limits their use in doctor-patient communication and clinical decision making. Here, we present the first large-scale clinical 3D morphable model, a machine-learning-based framework involving supervised learning for diagnostics, risk stratification, and treatment simulation. The model, trained and validated with 4,261 faces of healthy volunteers and orthognathic (jaw) surgery patients, diagnoses patients with 95.5% sensitivity and 95.2% specificity, and simulates surgical outcomes with a mean accuracy of 1.1 ± 0.3 mm. We demonstrate how this model could fully-automatically aid diagnosis and provide patient-specific treatment plans from a 3D scan alone, to help efficient clinical decision making and improve clinical understanding of face shape as a marker for primary and secondary surgery.

Download Full-text

Patient-Specific Computational Modeling of Upper Extremity Arteriovenous Fistula Creation: Its Feasibility to Support Clinical Decision-Making

PLoS ONE ◽

10.1371/journal.pone.0034491 ◽

2012 ◽

Vol 7 (4) ◽

pp. e34491 ◽

Cited By ~ 23

Author(s):

Aron S. Bode ◽

Wouter Huberts ◽

E. Marielle H. Bosboom ◽

Wilco Kroon ◽

Wim P. M. van der Linden ◽

...

Keyword(s):

Decision Making ◽

Computational Modeling ◽

Arteriovenous Fistula ◽

Upper Extremity ◽

Clinical Decision Making ◽

Clinical Decision ◽

Patient Specific

Download Full-text

A clinical decision-making mechanism for context-aware and patient-specific remote monitoring systems using the correlations of multiple vital signs

Computer Methods and Programs in Biomedicine ◽

10.1016/j.cmpb.2016.10.018 ◽

2017 ◽

Vol 139 ◽

pp. 1-16 ◽

Cited By ~ 16

Author(s):

Abdur Rahim Mohammad Forkan ◽

Ibrahim Khalil

Keyword(s):

Decision Making ◽

Remote Monitoring ◽

Clinical Decision Making ◽

Vital Signs ◽

Clinical Decision ◽

Patient Specific ◽

Monitoring Systems ◽

Context Aware ◽

Remote Monitoring Systems

Download Full-text

Treatment Options in Oncology

JCO Clinical Cancer Informatics ◽

10.1200/cci.18.00017 ◽

2018 ◽

pp. 1-10 ◽

Cited By ~ 8

Author(s):

Cédric M. Panje ◽

Markus Glatzer ◽

Charlotta Sirén ◽

Ludwig Plasswilm ◽

Paul M. Putora

Keyword(s):

Decision Making ◽

Clinical Decision Making ◽

Medical Condition ◽

Treatment Options ◽

Treatment Strategies ◽

Clinical Decision ◽

Vital Role ◽

Patient Specific ◽

Treatment Preferences ◽

Multiple Treatment

Multiple treatment strategies exist for many oncologic problems. In this review, we provide a summary of various reasons for the existence of multiple treatment options in oncology, including factors that concern the treating physician (eg, treatment preferences), environmental factors (eg, financial, regulatory, and scientific aspects), and individual patient-specific factors (eg, medical condition, preferences). We demonstrate the vital role of available treatment options and their origins for clinical decision making and patient communication. These aspects are particularly helpful in the process of shared decision making, which is increasingly favored in situations where there are multiple medically reasonable options.

Download Full-text

the Use of Tailored Feedback to Improve Hematologist Competence and Clinical Decision Making for the Treatment of Patients with Multiple Myeloma: a Novel Web-Based Educational Solution

Blood ◽

10.1182/blood.v128.22.3570.3570 ◽

2016 ◽

Vol 128 (22) ◽

pp. 3570-3570

Author(s):

Patricia Repetto ◽

Noopur S. Raje ◽

Sara R Fagerlie

Keyword(s):

Decision Making ◽

Multiple Myeloma ◽

Effect Size ◽

Clinical Decision Making ◽

Conflicts Of Interest ◽

Clinical Problem ◽

Clinical Decision ◽

Patient Specific ◽

Web Based ◽

Based Instruction

Abstract Introduction/Background: Recent advances in the understanding and treatment of multiple myeloma (MM) have led to improvements in patient management, including stratifying patients according to disease- and patient-specific risk factors, identifying appropriate patients for autologous stem cell transplantation, selecting treatment and incorporating new therapies into practice, and managing adverse effects. Materials and Methods: An online educational program using clinical problem-based instruction methodology was developed for hematologists and posted November 30, 2015 (http://www.medscape.org/viewarticle/853712). Each patient case included interactivity in the form of clinical decision questions and knowledge assessmentquestions. Tailored feedback and potential consequences in response to clinical decision questions was provided to each learner. Learners who answered a question incorrectly on the first attempt were provided feedback without revealing the correct answer and given the opportunity to answer the question again (second attempt) [Figure 1]. To determine measurable improvements in competence and clinical decision making, first and second attempt answer choices were evaluated for the clinical decision questions. Overall effect size was calculated using Cohen's d to show the magnitude and strength of the consequence-based feedback learning method. Data were collected through February 17, 2016. Results: A total of 404 hematologists participated in the activity during the study period, and responses from 129 (those who completed all clinical decision questions) were assessed. Responses to the 6 clinical decision questions show that a range of 28% to 67% of learners answered a question correctly on the first attempt. After receiving feedback specific to each incorrect answer, there was an overall 71% relative improvement in hematologists/oncologists who answered a question correctly on the second attempt [Figure 2]. The consequence-based feedback had a large impact on hematologists' ability to make clinical decisions correctly as demonstrated by the overall effect size of d=.94. Conclusions: This study demonstrated the success of a web-based CME activity with tailored feedback to learners' responses. Participation in this activity led to large improvements in clinician knowledge and the ability to select appropriate evidence- based practice choices, which may ultimately improve outcomes of patients being treated for multiple myeloma. Disclosures No relevant conflicts of interest to declare.

Download Full-text