Wui Ip, MD is a pediatrician and physician informaticist. He is interested in applying machine learning to support clinical decision making and improve patient care.
- Clinical Informatics
- Application of AI in Medicine
Clinical Assistant Professor, Pediatrics
Honors & Awards
EdTech Innovations Minigrant, Stanford University School of Medicine (2018)
Teaching and Mentoring Academy Grant, Stanford University (2018)
Community Service Award, Harvard Medical School (2010)
Chancellor's Award for Student Excellence, State University of New York (2007)
Commencement Student Speaker, SUNY Binghamton (2007)
Phi Beta Kappa, Phi Beta Kappa (2005)
Boards, Advisory Committees, Professional Organizations
Fellows' Council Representative, Department of Pediatrics, Stanford University (2019 - 2020)
Member, American Medical Informatics Association (2019 - Present)
Board Certification: American Board of Preventive Medicine, Clinical Informatics (2022)
Fellowship: Stanford University Clinical Informatics Fellowship (2021) CA
Board Certification: American Board of Pediatrics, Pediatrics (2016)
Residency: UCSF Pediatric Residency (2016) CA
Medical Education: Harvard Medical School (2013) MA
Fellowship, Stanford University, Clinical Informatics
Fellowship, Stanford Biodesign Faculty Fellow (2018)
Residency, UCSF, Pediatrics (2016)
MD, Harvard Medical School (2013)
Community and International Work
Medical Students Cultural Exchange, Taiwan
Cross-Cultural Education in Medicine
Mackay Medical College
Opportunities for Student Involvement
Additional Clinical Info
Technology and Education
Clinical outcome prediction using observational supervision with electronic health records and audit logs.
Journal of biomedical informatics
Audit logs in electronic health record (EHR) systems capture interactions of providers with clinical data. We determine if machine learning (ML) models trained using audit logs in conjunction with clinical data ("observational supervision") outperform ML models trained using clinical data alone in clinical outcome prediction tasks, and whether they are more robust to temporal distribution shifts in the data.Using clinical and audit log data from our hospital, we trained and evaluated various ML models including logistic regression, support vector machine (SVM) classifiers, neural networks, random forests, and gradient boosted machines (GBMs) on clinical EHR data, with and without audit logs for two clinical outcome prediction tasks: major adverse kidney events within 120 days of ICU admission (MAKE-120) in acute kidney injury (AKI) patients and 30-day readmission in acute stroke patients. We further tested the best performing models using patient data acquired during different time-intervals to evaluate the impact of temporal distribution shifts on model performance.Performance generally improved for all models when trained with clinical EHR data and audit log data compared with those trained with only clinical EHR data, with GBMs tending to have the overall best performance GBMs trained with clinical EHR data and audit logs outperformed GBMs trained without audit logs in both clinical outcome prediction tasks: AUROC 0.88 (95% CI: 0.85-0.91) vs. 0.79 (95% CI: 0.77-0.81), respectively, for MAKE-120 prediction in AKI patients, and AUROC 0.74 (95% CI: 0.71-0.77) vs. 0.63 (95% CI: 0.62-0.64), respectively, for 30-day readmission prediction in acute stroke patients. The performance of GBM models trained using audit log and clinical data degraded less in later time-intervals than models trained using only clinical data.Observational supervision with audit logs improved the performance of ML models trained to predict important clinical outcomes in patients with AKI and acute stroke, and improved robustness to temporal distribution shifts.
View details for DOI 10.1016/j.jbi.2023.104522
View details for PubMedID 37827476
Clinical course of children 1-24 months old with positive urine cultures without pyuria.
To describe the clinical course of children with positive urine cultures without pyuria who were not given antibiotics initially, identify predictors of subsequent antibiotic treatment, and evaluate the association between subsequent treatment and urinary tract infection (UTI) within 30 days.We conducted a multicenter retrospective cohort study of children 1-24 months old who had positive urine cultures without pyuria and who were not started on antibiotics upon presentation to 3 healthcare systems from 2010-2021. Outcomes included clinical status at the time urine cultures resulted, escalation of care (emergency department visit or hospitalization) and subsequent antibiotic treatment within 7 days, and subsequent UTI within 30 days of presentation.Of 202 included children, 61% were female and median age was 9 months. Of 151 patients with clinical status information when cultures resulted, 107 (70.8%, 95% confidence interval [CI] 62.9-77.9%) were improved. Two of 202 children (1.0%, 95% CI 0.2-4.0%) experienced care escalation. Antibiotics were started in 142 (82.2%) children and treatment was associated with prior UTI (risk ratio [RR] 1.20, 95% CI 1.15-1.26) and lack of improvement (RR 1.22, 95% CI 1.13-1.33). Subsequent UTI was diagnosed in 2 of 164 (1.2%, 95% CI 0.1-4.3%) treated and 0 of 36 (0%, 95% CI 0-9.7%) untreated children.Seventy percent of children with positive urine cultures without pyuria improved before starting antibiotics, however >80% were ultimately treated. Future research should study the impact of diagnostic stewardship interventions and various urine testing strategies to optimize management of children evaluated for UTI.Seventy percent of children with positive urine cultures without pyuria improved before starting antibiotics, however >80% were ultimately treated. Future research should study the impact of diagnostic stewardship interventions to optimize management of children evaluated for UTI.
View details for DOI 10.1016/j.acap.2023.06.023
View details for PubMedID 37354950
A Data-Driven Algorithm to Recommend Initial Clinical Workup for Outpatient Specialty Referral: Algorithm Development and Validation Using Electronic Health Record Data and Expert Surveys.
JMIR medical informatics
2022; 10 (3): e30104
BACKGROUND: Millions of people have limited access to specialty care. The problem is exacerbated by ineffective specialty visits due to incomplete prereferral workup, leading to delays in diagnosis and treatment. Existing processes to guide prereferral diagnostic workup are labor-intensive (ie, building a consensus guideline between primary care doctors and specialists) and require the availability of the specialists (ie, electronic consultation).OBJECTIVE: Using pediatric endocrinology as an example, we develop a recommender algorithm to anticipate patients' initial workup needs at the time of specialty referral and compare it to a reference benchmark using the most common workup orders. We also evaluate the clinical appropriateness of the algorithm recommendations.METHODS: Electronic health record data were extracted from 3424 pediatric patients with new outpatient endocrinology referrals at an academic institution from 2015 to 2020. Using item co-occurrence statistics, we predicted the initial workup orders that would be entered by specialists and assessed the recommender's performance in a holdout data set based on what the specialists actually ordered. We surveyed endocrinologists to assess the clinical appropriateness of the predicted orders and to understand the initial workup process.RESULTS: Specialists (n=12) indicated that <50% of new patient referrals arrive with complete initial workup for common referral reasons. The algorithm achieved an area under the receiver operating characteristic curve of 0.95 (95% CI 0.95-0.96). Compared to a reference benchmark using the most common orders, precision and recall improved from 37% to 48% (P<.001) and from 27% to 39% (P<.001) for the top 4 recommendations, respectively. The top 4 recommendations generated for common referral conditions (abnormal thyroid studies, obesity, amenorrhea) were considered clinically appropriate the majority of the time by specialists surveyed and practice guidelines reviewed.CONCLUSIONS: An item association-based recommender algorithm can predict appropriate specialists' workup orders with high discriminatory accuracy. This could support future clinical decision support tools to increase effectiveness and access to specialty referrals. Our study demonstrates important first steps toward a data-driven paradigm for outpatient specialty consultation with a tier of automated recommendations that proactively enable initial workup that would otherwise be delayed by awaiting an in-person visit.
View details for DOI 10.2196/30104
View details for PubMedID 35238788
Impact of telemedicine on clinical practice patterns for patients with chest pain in the emergency department.
International journal of medical informatics
2022; 161: 104726
BACKGROUND: The outbreak of the COVID-19 pandemic has led to the rapid adoption of novel telemedicine programs within the emergency department (ED) to minimize provider exposure and conserve personal protective equipment (PPE). In this study, we sought to assess how the adoption of telemedicine in the ED impacted clinical order patterns for patients with chest pain. We hypothesize that clinicians would rely more on imaging and laboratory workup for patients receiving telemedicine due to limitation in physical exams.METHODS: A single-center, retrospective, propensity score matched study was designed for patients presenting with chest pain at an ED. The study period was defined between April 1st, 2020 and September 30th, 2020. The frequency of the most frequent lab, imaging, and medication orders were compared. In addition, poisson regression analysis was performed to compare the overall number of orders between the two groups.RESULTS: 455 patients with chest pain who received telemedicine were matched to 455 similar patients without telemedicine with standardized mean difference<0.1 for all matched covariates. The proportion of frequent lab, imaging, and medication orders were similar between the two groups. However, telemedicine patients received more orders overall (RR, 1.19, 95% CI, 1.11, 1.28, p-value<0.001) as well as more imaging, lab, and nursing orders. The number of medication orders between the two groups remained similar.CONCLUSIONS: Frequent labs, imaging, and medications were ordered in similar proportions between the two cohorts. However, telemedicine patients had more orders placed overall. This study is an important objective assessment of the impact that telemedicine has upon clinical practice patterns and can guide future telemedicine implementation after the COVID-19 pandemic.
View details for DOI 10.1016/j.ijmedinf.2022.104726
View details for PubMedID 35228006
Ensuring Adolescent Patient Portal Confidentiality in the Age of the Cures Act Final Rule.
The Journal of adolescent health : official publication of the Society for Adolescent Medicine
PURPOSE: Managing confidential adolescent health information in patient portals presents unique challenges. Adolescent patients and guardians electronically access medical records and communicate with providers via portals. In confidential matters like sexual health, ensuring confidentiality is crucial. A key aspect of confidential portals is ensuring that the account is registered to and utilized by the intended user. Inappropriately registered or guardian-accessed adolescent portal accounts may lead to confidentiality breaches.METHODS: We used a quality improvement framework to develop screening methodologies to flag guardian-accessible accounts. Accounts of patients aged 12-17 were flagged via manual review of account emails and natural language processing of portal messages. We implemented a reconciliation program to correct affected accounts' registered email. Clinics were notified about sign-up errors and educated on sign-up workflow. An electronic alert was created to check the adolescent's email prior to account activation.RESULTS: After initial screening, 2,307 of 3,701 (62%) adolescent accounts were flagged as registered with a guardian's email. Those accounts were notified to resolve their logins. After five notifications over 8 weeks, 266 of 2,307 accounts (12%) were corrected; the remaining 2,041 (88%) were deactivated.CONCLUSIONS: The finding that 62% of adolescent portal accounts were used/accessed by guardians has significant confidentiality implications. In the context of the Cures Act Final Rule and increased information sharing, our institution's experience with ensuring appropriate access to adolescent portal accounts is necessary, timely, and relevant. This study highlights ways to improve patient portal confidentiality and prompts institutions caring for adolescents to review their systems and processes.
View details for DOI 10.1016/j.jadohealth.2021.09.009
View details for PubMedID 34666956
Assessment of Prevalence of Adolescent Patient Portal Account Access by Guardians.
JAMA network open
2021; 4 (9): e2124733
Importance: Patient portals can be configured to allow confidential communication for adolescents' sensitive health care information. Guardian access of adolescent patient portal accounts could compromise adolescents' confidentiality.Objective: To estimate the prevalence of guardian access to adolescent patient portals at 3 academic children's hospitals.Design, Setting, and Participants: A cross-sectional study to estimate the prevalence of guardian access to adolescent patient portal accounts was conducted at 3 academic children's hospitals. Adolescent patients (aged 13-18 years) with access to their patient portal account with at least 1 outbound message from their portal during the study period were included. A rule-based natural language processing algorithm was used to analyze all portal messages from June 1, 2014, to February 28, 2020, and identify any message sent by guardians. The sensitivity and specificity of the algorithm at each institution was estimated through manual review of a stratified subsample of patient accounts. The overall proportion of accounts with guardian access was estimated after correcting for the sensitivity and specificity of the natural language processing algorithm.Exposures: Use of patient portal.Main Outcome and Measures: Percentage of adolescent portal accounts indicating guardian access.Results: A total of 3429 eligible adolescent accounts containing 25 642 messages across 3 institutions were analyzed. A total of 1797 adolescents (52%) were female and mean (SD) age was 15.6 (1.6) years. The percentage of adolescent portal accounts with apparent guardian access ranged from 52% to 57% across the 3 institutions. After correcting for the sensitivity and specificity of the algorithm based on manual review of 200 accounts per institution, an estimated 64% (95% CI, 59%-69%) to 76% (95% CI, 73%-88%) of accounts with outbound messages were accessed by guardians across the 3 institutions.Conclusions and Relevance: In this study, more than half of adolescent accounts with outbound messages were estimated to have been accessed by guardians at least once. These findings have implications for health systems intending to rely on separate adolescent accounts to protect adolescent confidentiality.
View details for DOI 10.1001/jamanetworkopen.2021.24733
View details for PubMedID 34529064
- Factors Influencing Business Analytics Solutions and Views on Business Problems DATA 2021; 6 (8)
Day of Illness and Outcomes in Bronchiolitis Hospitalizations.
Bronchiolitis is often described to follow an expected clinical trajectory, with a peak in severity between days 3 and 5. This predicted trajectory may influence anticipatory guidance and clinical decision-making. We aimed to determine the association between day of illness at admission and outcomes, including hospital length of stay, receipt of positive-pressure ventilation, and total cough duration.We compiled data from 2 multicenter prospective studies involving bronchiolitis hospitalizations in patients <2 years. Patients were excluded for complex conditions. We assessed total cough duration via weekly postdischarge phone calls. We used mixed-effects multivariable regression models to test associations between day of illness and outcomes, with adjustment for age, sex, insurance (government versus nongovernment), race, and ethnicity.The median (interquartile range) day of illness at admission for 746 patients was 4 (2-5) days. Day of illness at admission was not associated with length of stay (coefficient 0.01 days, 95% confidence interval [CI]: -0.05 to 0.08 days), positive-pressure ventilation (adjusted odds ratio: 1.0, 95% CI: 0.9 to 1.1), or total cough duration (coefficient 0.33 days, 95% CI: -0.01 to 0.67 days). Additionally, there was no significant difference in day of illness at discharge in readmitted versus nonreadmitted patients (5.9 vs 6.4 days, P = .54). The median cough duration postdischarge was 6 days, with 65 (14.3%) patients experiencing cough for 14+ days.We found no associations between day of illness at admission and outcomes in bronchiolitis hospitalizations. Practitioners should exercise caution when making clinical decisions or providing anticipatory guidance based on symptom duration.
View details for DOI 10.1542/peds.2020-1537
View details for PubMedID 33093138
Evaluating Subcriticality during the Ebola Epidemic in West Africa
2015; 10 (10)
The 2014-2015 Ebola outbreak is the largest and most widespread to date. In order to estimate ongoing transmission in the affected countries, we estimated the weekly average number of secondary cases caused by one individual infected with Ebola throughout the infectious period for each affected West African country using a stochastic hidden Markov model fitted to case data from the World Health Organization. If the average number of infections caused by one Ebola infection is less than 1.0, the epidemic is subcritical and cannot sustain itself. The epidemics in Liberia and Sierra Leone have approached subcriticality at some point during the epidemic; the epidemic in Guinea is ongoing with no evidence that it is subcritical. Response efforts to control the epidemic should continue in order to eliminate Ebola cases in West Africa.
View details for DOI 10.1371/journal.pone.0140651
View details for Web of Science ID 000363028100058
View details for PubMedID 26484544
Interparticle interactions in glutathione mediated assembly of gold nanoparticles
2008; 24 (16): 8857-8863
The understanding of the detailed molecular interactions between (GSH) glutathione molecules in the assembly of metal nanoparticles is important for the exploitation of the biological reactivity. We report herein results of an investigation of the assembly of gold nanoparticles mediated by glutathione and the disassembly under controlled conditions. The interparticle interactions and reactivities were characterized by monitoring the evolution of the surface plasmon resonance band using the spectrophotometric method and the hydrodynamic sizes of the nanoparticle assemblies using the dynamic light scattering technique. The interparticle reactivity of glutathiones adsorbed on gold nanoparticles depends on the particle sizes and the ionic strength of the solution. Larger-sized particles were found to exhibit a higher degree of interparticle assembly than smaller-sized particles. The assembly-disassembly reversibility is shown to be highly dependent on pH and additives in the solution. The interactions of the negatively charged citrates surrounding the GSH monolayer on the particle surface were believed to produce more effective interparticle spatial and electrostatic isolation than the case of OH (-) groups surrounding the GSH monolayer. The results have provided new insights into the hydrogen-bonding character of the interparticle molecular interaction of glutathiones bound on gold nanoparticles. The fact that the interparticle hydrogen-bonding interactions in the assembly and disassembly processes can be finely tuned by pH and chemical means has implications to the exploitation of the glutathione-nanoparticle system in biological detection and biosensors.
View details for DOI 10.1021/la800970p
View details for Web of Science ID 000258377900069
View details for PubMedID 18642936
Contact-mediated cell-assisted cell proliferation in a model eukaryotic single-cell organism: An explanation for the lag phase in shaken cell culture
PHYSICAL REVIEW E
2008; 77 (4)
In cell culture, when cells are inoculated into fresh media, there can be a period of slow (or lag phase) growth followed by a transition to exponential growth. This period of slow growth is usually attributed to the cells' adaptation to a new environment. However, we argue that, based on observations of shaken suspension culture of Dictyostelium discoideum, a model single-cell eukaryote, this transition is due to a density effect. Attempts to demonstrate the existence of implicit cell signaling via long-range diffusible messengers (i.e., soluble growth factors) through cell-medium separation and microfluidic flow perturbation experiments produced negative results. This, in turn, led to the development of a signaling model based on direct cell-to-cell contacts. Employing a scaling argument for the collision rate due to fluid shear, we reasonably estimate the crossover density for the transition into the exponential phase and fit the observed growth kinetics.
View details for DOI 10.1103/PhysRevE.77.041905
View details for Web of Science ID 000255456900086
View details for PubMedID 18517654
Homocysteine-mediated reactivity and assembly of gold nanoparticles
2007; 23 (2): 826-833
This paper reports the findings of an investigation of the reactivity and assembly of gold nanoparticles mediated by homocysteine (Hcys), a thiol-containing amino acid found in plasma. The aim is to gain insight into the interparticle interaction and reactivity, which has potential application for the detection of thiol-containing amino acids. By monitoring the evolution of the surface plasmon resonance absorption and the dynamic light scattering of gold nanoparticles in the presence of Hcys, the assembly was shown to be dependent on the nature and concentration of the electrolytes, reflecting an effective screening of the diffuse layer around the initial citrate-capped nanoparticles that decreases the barrier to the Hcys adsorption onto the surface, and around the subsequent Hcys-capped nanoparticles that facilitate the zwitterion-type electrostatic interactions between amino acid groups of Hcys bound to different nanoparticles. A key element of the finding is that the interparticle zwitterion interaction of the Hcys-Au system is much stronger than the expectation for a simple Hcys or Au solution, a new phenomenon originating from the unique nanoscale interparticle interaction. The strength and reversibility of the interparticle zwitterion-type electrostatic interactions between amino acid groups are evidenced by the slow disassembly upon increasing pH at ambient temperatures and its acceleration at elevated temperature. These findings provide new insight into the precise control of interfacial interactions and reactivities between amino acids anchored to nanoparticles and have broad implications in the development of colorimetric nanoprobes for amino acids.
View details for DOI 10.1021/la062334t
View details for Web of Science ID 000243338500069
View details for PubMedID 17209640
Adsorption of cyanine dyes on gold nanoparticles and formation of J-aggregates in the nanoparticle assembly
JOURNAL OF PHYSICAL CHEMISTRY B
2006; 110 (13): 6673-6682
This paper describes the results of an investigation of the interparticle interactions and reactivities in the assembly of gold nanoparticles mediated by cyanine dyes. The combination of the positively charged indolenine cyanine dyes and the negatively charged gold nanoparticles is shown to form a J-aggregate bridged assembly of nanoparticles, in addition to hydrophobic interparticle and electrostatic dye-particle interactions. Such interparticle interactions and reactivities are studied by probing the absorption of J-aggregates and fluorescence from the dyes and the surface plasmon resonance absorption from the nanoparticles. The J-aggregation of the dyes adsorbed on the nanoparticles is shown to play an important role in the assembly of nanoparticles. The spectral evolution of the J-band of the dyes and the surface plasmon resonance band of the nanoparticles was found to be sensitive to the nature of the charge and the structure of the dyes. The fluorescence quenching for the dyes was shown to be quantitatively related to the surface coverage of the dyes on the nanocrystal surfaces. These findings have provided important information for assessing a two-step process involving a rapid adsorption of the dyes on the nanoparticles and a subsequent assembly of the nanoparticles involving a combination of interparticle J-aggregation and hydrophobic interactions of the adsorbed dyes. The results are discussed in terms of the structural effects of the dyes, and the interparticle molecular interactions and reactivities, which provide important physical and chemical insights into the design of dye-nanoparticle structured functional nanomaterials.
View details for DOI 10.1021/jp057584h
View details for Web of Science ID 000236523100040
View details for PubMedID 16570972