Christoph Kern

Predicting Visual Acuity by Using Machine Learning in Patients Treated for Neovascular Age-Related Macular Degeneration

PURPOSEnTo predict, by using machine learning, visual acuity (VA) at 3 and 12 months in patients with neovascular age-related macular degeneration (AMD) after initial upload of 3 anti-vascular endothelial growth factor (VEGF) injections.nnnDESIGNnDatabase study.nnnPARTICIPANTSnFor the 3-month VA forecast, 653 patients (379 female) with 738 eyes and an average age of 74.1 years were included. The baseline VA before the first injection was 0.54 logarithm of the minimum angle of resolution (logMAR) (±0.39). A total of 456 of these patients (270 female, 508 eyes, average age: 74.2 years) had sufficient follow-up data to be included for a 12-month VA prediction. The baseline VA before the first injection was 0.56 logMAR (±0.42).nnnMETHODSnFive different machine-learning algorithms (AdaBoost.R2, Gradient Boosting, Random Forests, Extremely Randomized Trees, and Lasso) were used to predict VA in patients with neovascular AMD after treatment with 3 anti-VEGF injections. Clinical data features came from a data warehouse (DW) containing electronic medical records (41 features, e.g., VA) and measurement features from OCT (124 features, e.g., central retinal thickness). The VA of patient eyes excluded from machine learning was predicted and compared with the ground truth, namely, the actual VA of these patients as recorded in the DW.nnnMAIN OUTCOME MEASURESnDifference in logMAR VA after 3 and 12 months upload phase between prediction and ground truth as defined.nnnRESULTSnFor the 3-month VA forecast, the difference between the prediction and ground truth was between 0.11 logMAR (5.5 letters) mean absolute error (MAE)/0.14 logMAR (7 letters) root mean square error (RMSE) and 0.18 logMAR (9 letters) MAE/0.2 logMAR (10 letters) RMSE. For the 12-month VA forecast, the difference between the prediction and ground truth was between 0.16 logMAR (8 letters) MAE/0.2 logMAR (10 letters) RMSE and 0.22 logMAR (11 letters) MAE/0.26 logMAR (13 letters) RMSE. The best performing algorithm was the Lasso protocol.nnnCONCLUSIONSnMachine learning allowed VA to be predicted for 3 months with a comparable result to VA measurement reliability. For a forecast after 12 months of therapy, VA prediction may help to encourage patients adhering to intravitreal therapy.

Using Electronic Health Records to Build an Ophthalmologic Data Warehouse and Visualize Patients' Data

PURPOSEnTo develop a near-real-time data warehouse (DW) in an academic ophthalmologic center to gain scientific use of increasing digital data from electronic medical records (EMR) and diagnostic devices.nnnDESIGNnDatabase development.nnnMETHODSnSpecific macular clinic user interfaces within the institutional hospital information system were created. Orders for imaging modalities were sent by an EMR-linked picture-archiving and communications system to the respective devices. All data of 325 767 patients since 2002 were gathered in a DW running on an SQL database. A data discovery tool was developed. An exemplary search for patients with age-related macular degeneration, performed cataract surgery, and at least 10 intravitreal (excluding bevacizumab) injections was conducted.nnnRESULTSnData related to those patients (3 142 204 diagnoses [including diagnoses from other fields of medicine], 720 721 procedures [eg, surgery], and 45 416 intravitreal injections) were stored, including 81 274 optical coherence tomography measurements. A web-based browsing tool was successfully developed for data visualization and filtering data by several linked criteria, for example, minimum number of intravitreal injections of a specific drug and visual acuity interval. The exemplary search identified 450 patients with 516 eyes meeting all criteria.nnnCONCLUSIONSnA DW was successfully implemented in an ophthalmologic academic environment to support and facilitate research by using increasing EMR and measurement data. The identification of eligible patients for studies was simplified. In future, software for decision support can be developed based on the DW and its structured data. The improved classification of diseases and semiautomatic validation of data via machine learning are warranted.

Comparison of visual outcomes, alignment accuracy, and surgical time between 2 methods of corneal marking for toric intraocular lens implantation

PURPOSEnTo compare the efficacy of a computer-assisted marker system for toric intraocular lenses (IOLs) (Callisto Eye System) with manual marking techniques.nnnSETTINGnUniversity Eye Hospital Munich, Ludwig-Maximilians-University, Munich, Germany.nnnDESIGNnProspective case series.nnnMETHODSnThis study included patients having cataract surgery with implantation of a toric IOL (Torbi 709xa0M). They were randomly assigned to 1 of 2 groups based on the marking system used, manual or digital. Patients were included if they had age-related cataract and a regular corneal astigmatism of 1.25 diopters (D) or higher. Visual and refractive outcomes as well as rotational stability were evaluated. Vector analysis was performed to evaluate total astigmatic changes.nnnRESULTSnThe study comprised 57 eyes of 29 patients; there were 28 eyes in the manual group and 29 eyes in the digital group. The mean toric IOL misalignment was significantly lowerxa0in the digital group than in the manual group (2.0 degreesxa0±xa01.86 [SD] versus 3.4xa0±xa02.37 degrees; Pxa0=xa0.026). The mean deviation from the target induced astigmatism was significantly lower in the digital group (0.10xa0±xa00.08 D versus 0.22xa0±xa00.14 D; Pxa0=xa0.008). During surgery, the mean toric IOL alignment time was significantly shorter in the digital groupxa0(37.2xa0±xa011.9xa0seconds versus 59.4xa0±xa015.3xa0seconds; Pxa0=xa0.003). The mean overall time required to perform the surgeryxa0was significantly shorter in the digital group (727.2xa0±xa0198.4xa0seconds versus 1110.0xa0±xa0382.2xa0seconds; Pxa0<xa0.001).nnnCONCLUSIONSnA digital tracking approach for toric IOL alignment was efficient and safe to improve refractive outcomes. Furthermore, image-guided surgery helped streamline the workflow in refractive cataract surgery.

Smart eye data

ZusammenfassungHintergrundUnter Smart Data versteht man die intelligente Sammelplanung, Umsetzung und Evaluation großer Datenmengen. Dies trifft auch auf die Augenheilkunde zu, da mehr und mehr Daten digital anfallen. Einen zusätzlichen Erkenntnisgewinn sowie Möglichkeiten der personalisierten Therapie erwartet man sich aus der Kombination der Messdaten mit klinischen Werten aus einer digitalen Akte.Ziel der ArbeitIm Rahmen dieser Arbeit soll gezeigt werden, wie man die Daten aus Messgeräten sowie einer digitalen Akte in einem „Data Warehouse“ (DW) zusammenführen kann.Material und MethodenEs wurde eine Erweiterung des klinischen Informationssystems (Krankenhausinformationssystem [KIS]) um Eingabemasken für die Augenheilkunde durchgeführt, sowie Arztbriefe wurden exportiert. Diese Daten werden nächtlich in ein DW importiert. Die Anbindung von Diagnostikgerät an das KIS per HL7-Schnittstelle erfolgte über ein Bildarchivsystem. Der Export von Messdaten in das DW geschah über selbst entwickelte Programme. Zur Datenanalyse wurde eine Software für augenärztliche Bedürfnisse modifiziert.ErgebnisseEs wurden 12xa0Eingabemasken innerhalb des KIS geschaffen sowie der Inhalt von 32.234xa0Arztbriefen importiert. Es konnten 23xa0Diagnostikgeräte an das Bildverwaltungssystem angeschlossen werden und 85.114xa0OCT (optische Kohärenztomographie)-Datensätze, 19.098xa0IOLMaster-Datensätze sowie 5425xa0Pentacam-Datensätze in das DW mit über 300.000xa0Patienten importiert werden. Zur Datenanalyse wurde eine Auswertungssoftware entwickelt, die Filtermöglichkeiten bietet.DiskussionDurch den Aufbau eines DW steht eine Grundlage für klinische und epidemiologische Auswertungen zur Verfügung. In Zukunft kann der Datenbestand genutzt werden, um Entscheidungsfindungssysteme sowie Anwendungen für personalisierte Medizin zu entwickeln.AbstractBackgroundSmart Data means intelligent data accumulation and the evaluation of large data sets. This is particularly important in ophthalmology as more and more data are being created. Increasing knowledge and personalized therapies are expected by combining clinical data from electronic health records (EHR) with measurement data.ObjectiveIn this study we investigated the possibilities to consolidate data from measurement devices and clinical data in a data warehouse (DW).Material and methodsAn EHR was adjusted to the needs of ophthalmology and the contents of referral letters were extracted. The data were imported into a DW overnight. Measuring devices were connected to the EHR by an HL7 standard interface and the use of a picture archiving and communications system (PACS). Data were exported from the review software using a self-developed software. For data analysis the software was modified to the specific requirements of ophthalmology.ResultsIn the EHR 12 graphical user interfaces were created and the data from 32,234 referral letters were extracted. A total of 23 diagnostic devices could be linked to the PACS and 85,114 optical coherence tomography (OCT) scans, 19,098 measurements from IOLMaster as well as 5,425 pentacam examinations were imported into the DW including over 300,000 patients. Data discovery software was modified providing filtering methods.ConclusionBy building a DW a foundation for clinical and epidemiological studies could be implemented. In the future, decision support systems and strategies for personalized therapies can be based on such a database.

[Smart eye data : Development of a foundation for medical research using Smart Data applications].

ZusammenfassungHintergrundUnter Smart Data versteht man die intelligente Sammelplanung, Umsetzung und Evaluation großer Datenmengen. Dies trifft auch auf die Augenheilkunde zu, da mehr und mehr Daten digital anfallen. Einen zusätzlichen Erkenntnisgewinn sowie Möglichkeiten der personalisierten Therapie erwartet man sich aus der Kombination der Messdaten mit klinischen Werten aus einer digitalen Akte.Ziel der ArbeitIm Rahmen dieser Arbeit soll gezeigt werden, wie man die Daten aus Messgeräten sowie einer digitalen Akte in einem „Data Warehouse“ (DW) zusammenführen kann.Material und MethodenEs wurde eine Erweiterung des klinischen Informationssystems (Krankenhausinformationssystem [KIS]) um Eingabemasken für die Augenheilkunde durchgeführt, sowie Arztbriefe wurden exportiert. Diese Daten werden nächtlich in ein DW importiert. Die Anbindung von Diagnostikgerät an das KIS per HL7-Schnittstelle erfolgte über ein Bildarchivsystem. Der Export von Messdaten in das DW geschah über selbst entwickelte Programme. Zur Datenanalyse wurde eine Software für augenärztliche Bedürfnisse modifiziert.ErgebnisseEs wurden 12xa0Eingabemasken innerhalb des KIS geschaffen sowie der Inhalt von 32.234xa0Arztbriefen importiert. Es konnten 23xa0Diagnostikgeräte an das Bildverwaltungssystem angeschlossen werden und 85.114xa0OCT (optische Kohärenztomographie)-Datensätze, 19.098xa0IOLMaster-Datensätze sowie 5425xa0Pentacam-Datensätze in das DW mit über 300.000xa0Patienten importiert werden. Zur Datenanalyse wurde eine Auswertungssoftware entwickelt, die Filtermöglichkeiten bietet.DiskussionDurch den Aufbau eines DW steht eine Grundlage für klinische und epidemiologische Auswertungen zur Verfügung. In Zukunft kann der Datenbestand genutzt werden, um Entscheidungsfindungssysteme sowie Anwendungen für personalisierte Medizin zu entwickeln.AbstractBackgroundSmart Data means intelligent data accumulation and the evaluation of large data sets. This is particularly important in ophthalmology as more and more data are being created. Increasing knowledge and personalized therapies are expected by combining clinical data from electronic health records (EHR) with measurement data.ObjectiveIn this study we investigated the possibilities to consolidate data from measurement devices and clinical data in a data warehouse (DW).Material and methodsAn EHR was adjusted to the needs of ophthalmology and the contents of referral letters were extracted. The data were imported into a DW overnight. Measuring devices were connected to the EHR by an HL7 standard interface and the use of a picture archiving and communications system (PACS). Data were exported from the review software using a self-developed software. For data analysis the software was modified to the specific requirements of ophthalmology.ResultsIn the EHR 12 graphical user interfaces were created and the data from 32,234 referral letters were extracted. A total of 23 diagnostic devices could be linked to the PACS and 85,114 optical coherence tomography (OCT) scans, 19,098 measurements from IOLMaster as well as 5,425 pentacam examinations were imported into the DW including over 300,000 patients. Data discovery software was modified providing filtering methods.ConclusionBy building a DW a foundation for clinical and epidemiological studies could be implemented. In the future, decision support systems and strategies for personalized therapies can be based on such a database.

Assessing novel prognostic serum biomarkers in advanced pancreatic cancer: the role of CYFRA 21-1, serum amyloid A, haptoglobin, and 25-OH vitamin D3

The present prospective single-center study investigated the prognostic role of novel serum biomarkers in advanced pancreatic cancer (PC). Patients (pts) with locally advanced or metastatic PC treated with first-line palliative chemotherapy were included. Among others, the serum markers CYFRA 21-1, haptoglobin, serum-amyloid A (SAA), and 25-OH vitamin D3 were determined at baseline and categorized by pre-defined cut-offs [median values (MV), upper limits of normal (ULN), lower limits of normal (LLN), or the natural logarithm (ln)] and correlated with overall survival (OS). Among the 59 pts included, pre-treatment CYFRA 21-1 levels showed a strong correlation with OS independent of the applied cut-off (MV 4.9xa0ng/ml—14.2 vs. 4.2 months, HR 0.18, pu2009=u20090.001; ULN 3.3xa0ng/ml—14.2 vs. 4.4 months, HR 0.28, pu2009=u20090.003; [ln] CYFRA 21-1—HR 0.77, pu2009=u20090.013). Lower values of haptoglobin were additionally associated with an improvement in OS (categorized by LLN of 2.05xa0g/l—10.4 vs. 5.5 months, HR 0.46, pu2009=u20090.023; [ln] haptoglobin—HR 0.51, pu2009=u20090.036). Pts with baseline SAA values below the MV of 22xa0mg/l also had a prolonged OS (10.4 vs. 5.0 months, HR 0.47, pu2009=u20090.036). For 25-OH vitamin D3 levels, no significant correlation with OS was found. In multivariate analyses, pre-treatment CYFRA 21-1 levels (categorized by MV—HR 0.15, pu2009=u20090.032) as well as [ln] haptoglobin (HR 0.30, pu2009=u20090.006) retained their independent prognostic significance for OS. CYFRA 21-1, haptoglobin, and SAA might provide useful prognostic information in advanced PC. An external multicenter validation of these results is necessary.

Comparison of Two Toric IOL Calculation Methods

Purpose To compare two calculators for toric intraocular lens (IOL) calculation and to evaluate the prediction of refractive outcome. Methods Sixty-four eyes of forty-five patients underwent cataract surgery followed by implantation of a toric intraocular lens (Zeiss Torbi 709u2009M) calculated by a standard industry calculator using front keratometry values. Prediction error, median absolute error, and refractive astigmatism error were evaluated for the standard calculator. The predicted postoperative refraction and toric lens power values were evaluated and compared after postoperative recalculation using the Barrett calculator. Results We observed a significant undercorrection in the spherical equivalent (0.19u2009D) by using a standard calculator (p ≤ 0.05). According to the Baylor nomogram and the refractive influence of posterior corneal astigmatism (PCA), undercorrection of the cylinder was lower for patients with WTR astigmatism, because of the tendency of overcorrection. An advantage of less residual postoperative SE, sphere, and cylinder for the Barrett calculator was observed when retrospectively comparing the calculated predicted postoperative refraction between calculators (p ≤ 0.01). Conclusion Consideration of only corneal front keratometric values for toric lens calculation may lead to postoperative undercorrection of astigmatism. The prediction of postoperative refractive outcome can be improved by using appropriate methods of adjustment in order to take PCA into account.

Modern Corneal Eye-Banking Using a Software-Based IT Management Solution

Background Increasing government legislation and regulations in manufacturing have led to additional documentation regarding the pharmaceutical product requirements of corneal grafts in the European Union. The aim of this project was to develop a software within a hospital information system (HIS) to support the documentation process, to improve the management of the patient waiting list and to increase informational flow between the clinic and eye bank. Materials and Methods After an analysis of the current documentation process, a new workflow and software were implemented in our electronic health record (EHR) system. Results The software takes over most of the documentation and reduces the time required for record keeping. It guarantees real-time tracing of all steps during human corneal tissue processing from the start of production until allocation during surgery and includes follow-up within the HIS. Moreover, listing of the patient for surgery as well as waiting list management takes place in the same system. Conclusion The new software for corneal eye banking supports the whole process chain by taking over both most of the required documentation and the management of the transplant waiting list. It may provide a standardized IT-based solution for German eye banks working within the same HIS.

Differences in corneal clinical findings after standard and accelerated cross-linking in patients with progressive keratoconus

BackgroundThe purpose of this study was to identify differences in clinical corneal findings after standard and accelerated epithelial off cross-linking (CXL) during a long-term follow-up.MethodsTwo hundred forty-one patients (184 male) were included in this monocentric, retrospective, non-randomized and unmasked study. One hundred forty-eight eyes were treated with the accelerated protocol and 138 with the standard protocol with epithelial off CXL, if diagnosed with keratoconus and a progression in Kmax of more than one dioptre during the preceding 6 months, plus a minimal pachymetry measurement of 400xa0μm in keratometry (Pentacam, Oculus GmbH, Wetzlar, Germany). Exclusion criteria were previous surgery, other corneal conditions or age above 50xa0years. Follow-up time was 36xa0months with clinical examination and keratometry at every visit. Outcome measures were the observed rate of corneal changes, differences between treatment groups and correlation with keratometry measurements.ResultsIn patients with accelerated CXL, significantly more clear corneas were seen at three (pxa0=u20090.015) and six (pxa0=u20090.002) months after surgery than following the standard protocol. The rate of clear corneas dropped from 52.2% pre-operation (OP) to a minimum of 19.3% after 6xa0months in the standard protocol group compared with 50.7% clear corneas pre-OP and a minimum of 40.8% in the accelerated group. In the standard protocol group, more striae were found 3 months after intervention than in the accelerated group (pxa0=u20090.05).ConclusionsIn patients with accelerated CXL, fewer morphological corneal changes were observed than after conventional CXL. However, rarely, corneal changes persisted for a long time.

„Smart eye data“Smart eye data

ZusammenfassungHintergrundUnter Smart Data versteht man die intelligente Sammelplanung, Umsetzung und Evaluation großer Datenmengen. Dies trifft auch auf die Augenheilkunde zu, da mehr und mehr Daten digital anfallen. Einen zusätzlichen Erkenntnisgewinn sowie Möglichkeiten der personalisierten Therapie erwartet man sich aus der Kombination der Messdaten mit klinischen Werten aus einer digitalen Akte.Ziel der ArbeitIm Rahmen dieser Arbeit soll gezeigt werden, wie man die Daten aus Messgeräten sowie einer digitalen Akte in einem „Data Warehouse“ (DW) zusammenführen kann.Material und MethodenEs wurde eine Erweiterung des klinischen Informationssystems (Krankenhausinformationssystem [KIS]) um Eingabemasken für die Augenheilkunde durchgeführt, sowie Arztbriefe wurden exportiert. Diese Daten werden nächtlich in ein DW importiert. Die Anbindung von Diagnostikgerät an das KIS per HL7-Schnittstelle erfolgte über ein Bildarchivsystem. Der Export von Messdaten in das DW geschah über selbst entwickelte Programme. Zur Datenanalyse wurde eine Software für augenärztliche Bedürfnisse modifiziert.ErgebnisseEs wurden 12xa0Eingabemasken innerhalb des KIS geschaffen sowie der Inhalt von 32.234xa0Arztbriefen importiert. Es konnten 23xa0Diagnostikgeräte an das Bildverwaltungssystem angeschlossen werden und 85.114xa0OCT (optische Kohärenztomographie)-Datensätze, 19.098xa0IOLMaster-Datensätze sowie 5425xa0Pentacam-Datensätze in das DW mit über 300.000xa0Patienten importiert werden. Zur Datenanalyse wurde eine Auswertungssoftware entwickelt, die Filtermöglichkeiten bietet.DiskussionDurch den Aufbau eines DW steht eine Grundlage für klinische und epidemiologische Auswertungen zur Verfügung. In Zukunft kann der Datenbestand genutzt werden, um Entscheidungsfindungssysteme sowie Anwendungen für personalisierte Medizin zu entwickeln.AbstractBackgroundSmart Data means intelligent data accumulation and the evaluation of large data sets. This is particularly important in ophthalmology as more and more data are being created. Increasing knowledge and personalized therapies are expected by combining clinical data from electronic health records (EHR) with measurement data.ObjectiveIn this study we investigated the possibilities to consolidate data from measurement devices and clinical data in a data warehouse (DW).Material and methodsAn EHR was adjusted to the needs of ophthalmology and the contents of referral letters were extracted. The data were imported into a DW overnight. Measuring devices were connected to the EHR by an HL7 standard interface and the use of a picture archiving and communications system (PACS). Data were exported from the review software using a self-developed software. For data analysis the software was modified to the specific requirements of ophthalmology.ResultsIn the EHR 12 graphical user interfaces were created and the data from 32,234 referral letters were extracted. A total of 23 diagnostic devices could be linked to the PACS and 85,114 optical coherence tomography (OCT) scans, 19,098 measurements from IOLMaster as well as 5,425 pentacam examinations were imported into the DW including over 300,000 patients. Data discovery software was modified providing filtering methods.ConclusionBy building a DW a foundation for clinical and epidemiological studies could be implemented. In the future, decision support systems and strategies for personalized therapies can be based on such a database.