Search

Background: Medical coding is the process that converts clinical documentation into standard medical codes. Codes are used for several key purposes in a hospital (eg, insurance reimbursement and performance analysis); therefore, their optimization is crucial. With the rapid growth of natural language processing technologies, several solutions based on artificial intelligence have been proposed to aid in medical coding by automatically suggesting relevant codes for clinical documents. However, their effectiveness is still limited to simple cases, and it is not yet clear how much value they can bring in improving coding efficiency and accuracy., Objective: This study aimed to bring more efficiency to the coding process to improve the selection of codes by medical coders. To achieve this, we developed an innovative multimodal machine learning-based solution that, instead of predicting codes, detects the degree of coding complexity before coding is performed. The notion of coding complexity was used to better dispatch work among medical coders to eventually minimize errors and improve throughput., Methods: To train and evaluate our approach, we collected 2060 cases rated by coders in terms of coding complexity from 1 (simplest) to 4 (most complex). We asked 2 expert coders to rate 3.01% (62/2060) of the cases as the gold standard. The agreements between experts were used as benchmarks for model evaluation. A case contains both clinical text and patient metadata from the hospital electronic health record. We extracted both text features and metadata features, then concatenated and fed them into several machine learning models. Finally, we selected 2 models. The first used cross-validated training on 1751 cases and testing on 309 cases aiming to assess the predictive power of the proposed approach and its generalizability. The second model was trained on 1998 cases and tested on the gold standard to validate the best model performance against human benchmarks., Results: Our first model achieved a macro-F 1 -score of 0.51 and an accuracy of 0.59 on classifying the 4-scale complexity. The model distinguished well between the simple (combined complexity 1-2) and complex (combined complexity 3-4) cases with a macro-F 1 -score of 0.65 and an accuracy of 0.71. Our second model achieved 61% agreement with experts' ratings and a macro-F 1 -score of 0.62 on the gold standard, whereas the 2 experts had a 66% (41/62) agreement ratio with a macro-F 1 -score of 0.67., Conclusions: We propose a multimodal machine learning approach that leverages information from both clinical text and patient metadata to predict the complexity of coding a case in the precoding phase. By integrating this model into the hospital coding system, distribution of cases among coders can be done automatically with performance comparable with that of human expert coders, thus improving coding efficiency and accuracy at scale., (©He Ayu Xu, Bernard Maccari, Hervé Guillain, Julien Herzen, Fabio Agri, Jean Louis Raisaro. Originally published in JMIR Medical Informatics (https://medinform.jmir.org), 19.01.2023.)

Background: Candida species are among the most frequent causative agents of health care-associated bloodstream infections, with mortality >40% in critically ill patients. Specific populations of critically ill patients may present peculiar risk factors related to their reason for intensive care unit admission. The primary objective of the present study was to assess the predictors of candidemia after open heart surgery., Methods: This retrospective, matched case-control study was conducted in 8 Italian hospitals from 2009 to 2016. The primary study objective was to assess factors associated with the development of candidemia after open heart surgery., Results: Overall, 222 patients (74 cases and 148 controls) were included in the study. Candidemia developed at a median time (interquartile range) of 23 (14-36) days after surgery. In multivariable analysis, independent predictors of candidemia were New York Heart Association class III or IV (odds ratio [OR], 23.81; 95% CI, 5.73-98.95; P  < .001), previous therapy with carbapenems (OR, 8.87; 95% CI, 2.57-30.67; P  = .001), and previous therapy with fluoroquinolones (OR, 5.73; 95% CI, 1.61-20.41; P  = .007). Crude 30-day mortality of candidemia was 53% (39/74). Septic shock was independently associated with mortality in the multivariable model (OR, 5.64; 95% CI, 1.91-16.63; P  = .002). No association between prolonged cardiopulmonary bypass time and candidemia was observed in this study., Conclusions: Previous broad-spectrum antibiotic therapy and high NYHA class were independent predictors of candidemia in cardiac surgery patients with prolonged postoperative intensive care unit stay., (© The Author(s) 2020. Published by Oxford University Press on behalf of Infectious Diseases Society of America.)

Rationale: Patients receiving an allogeneic stem cell graft from cytomegalovirus (CMV) seronegative donors are particularly prone to CMV reactivation with a high risk of disease and mortality. Therefore we developed and manufactured a novel vaccine and initiated a clinical phase I trial with a CMV phosphoprotein 65 (CMVpp65)-derived peptide. Methods: Ten patients after allogeneic stem cell transplantation received four vaccinations at a biweekly interval. All patients were monitored for CMVpp65 antigenemia. Flow cytometry for CMV-specific CD8 + and γδ T cells as well as neutralizing anti-CMV antibodies were correlated to clinical parameters. Results: The vaccination was well tolerated. Seven of nine patients cleared CMVpp65 antigenemia after four vaccinations and are still free from antigenemia to this day. Two patients with CMV reactivation showed persisting CMV antigenemia. One patient received prophylactic vaccination and did not develop antigenemia. An increase of up to six-fold in frequency of both CMV-specific CD8 + T cells and/or Vδ2negative γδ T cells was detected. Titers of neutralizing antibodies increased up to the tenfold. Humoral and cellular immune responses correlated with clearance of CMV. Conclusion: In summary, CMVpp65 peptide vaccination for patients after allogeneic stem cell transplantation at high risk for CMV reactivation was safe, well tolerated and clinically encouraging. A study in solid-organ transplant patients is ongoing., Competing Interests: Competing Interests: J.K. is inventor of multiple patents dealing with γδ TCRs and co-founder as well as chief scientific officer (CSO) of Gadeta (http://www.gadeta.nl/). The other authors have declared that no competing interest exists.

Peptide vaccination constitutes a novel immunotherapeutical approach for the treatment of patients with solid tumors, lymphoma and leukemia. Moreover it might be of use in hematooncological patients for the prevention and therapy of infections like cytomegalovirus (CMV) reactivation due to immunosuppression. To meet good manufacturing practice (GMP) criteria, we introduce here a bio-assay to validate peptide vaccines for peptide content and bio-activity. As a paradigm for peptide vaccine preparation the immunogenic CMV peptide 495-503 NLVPMVATV lyophilisate was resolubilized in dimethyl sulfoxide, phosphate buffered saline and admixed with Montanide. Addition of different amounts of peptide (10-80 microg) to a mixed lymphocyte peptide culture (MLPC) resulted in the generation of interferon (IFN) gamma and granzyme B releasing CD8(+) CMV tetramer(+) T cells in a dose dependent manner. The combination of FACS and ELISPOT results allowed the definition of the peptide amount in a vaccine preparation. Storage at +/-4 degrees C over 24 h did not result in a significant change of the immunogenicity of the vaccine. In contrast, cryopreservation of the vaccine at -20 degrees C resulted in a loss of immunogenicity. Quantitation of tumor/viral antigen peptides admixed with adjuvants, such as incomplete Freund's adjuvant (IFA), is feasible through bio-assays as the modified ELISPOT/FACS assay described here, meeting GMP criteria for multi-center trials.

Chimerism analysis by DNA-based methods is a valuable diagnostic tool for monitoring engraftment and leukemic relapse after allogeneic BMT or PBPC transplantation (PBPCT). We investigated the chimerism after T-cell-depleted BMT (n = 32) in comparison with T-cell-depleted PBPCT (n = 39). BM grafts were T-cell depleted using the Campath-IgM antibody plus complement. For T-cell depletion of the PBPC grafts, a selection of CD34+ cells with or without a subsequent CD2/3 depletion was performed. In all patients, the T-cell dose of the transplant was < 10(6)/kg body weight. Between day 13 and day 120 after transplantation, chimerism analysis was done by RFLP or amplified fragment length polymorphism (PCR-AFLP), with a detection limit of 1%-5% recipient cells. In the BMT group, 8 of 32 (25%) patients showed a mixed chimerism, but only one graft rejection and no leukemic relapse occurred after a median follow-up of 41 (3-84) months. All patients with PBPCT revealed a complete chimerism of their granulocytes, and 38 of 39 patients showed complete chimerism of their lymphocytes. Follow-up time in these patients is 7 (2-21) months, with no graft rejection and two leukemic relapses. G-CSF-mobilized PBPC are superior to BM cells for full engraftment even after T-cell-depleted transplantation. The more relevant factor for developing complete chimerism seems to be the quantity and possibly the quality of the stem cells rather than the residual T-cell load of the graft. However, a mixed chimerism of the lymphocytes early after transplantation does not predict a higher rate of graft rejection or leukemic relapse.

A median dose of 11 (6-17) microg G-CSF per kg and day was given to 96 (49 female, 47 male) healthy family donors in order to mobilize and to collect peripheral blood progenitor cells (PBPC) for allogeneic transplantation. Donor age was 36 (17-76) years. The leukocytes of the donors increased to 46 (12-115) x 10(9)/l on days 4-6 of G-CSF treatment with a median of 71 (2-657) CD34+ cells per microl, respectively. Female and older donors seem to have a lower response to G-CSF. About 32% of the donors suffered from side-effects of G-CSF requiring analgetics. A total of 197 stem cell aphereses were performed using the COBE Spectra cell separator. Median apheresis time was 225 (118-300) min processing 11.8 (5.7-20) l blood, collecting 5.3 (1.7-14.9) x 10(10) nucleated cells and containing 0.7 (0.1-3.7)% CD34+ cells. Severe citrate toxicity occurred in 5% of the donors. Retransfusion of autologous platelets post apheresis was necessary in 16% of the donors because of a platelet count <80 x 10(9)/l. An insufficient number of stem cells was collected in four female donors due to a very poor response to G-CSF. In conclusion, the collection of allogeneic G-CSF-mobilized PBPC is safe and effective. One or two aphereses were sufficient in 91% of the donors to achieve >4 x 10(6) CD34+ cells per kg. In 4% of the donors an additional bone marrow harvest or the use of an alternative donor was necessary because of a poor mobilization.

Selection of CD34+ cells for autologous transplantation is increasingly being used to reduce potential tumor cell contamination of the autograft. Haematopoietic reconstitution in 40 patients after transplantation of CD34(+)-selected versus non-selected G-CSF-mobilized PBPC was compared and was almost identical in the two groups of patients. Delayed platelet engraftment was only observed in patients transplanted with a CD34+ cell dose of < 2.5 x 10(6)/kg body weight. It has to be shown whether the positive selection of CD34+ cells will improve the disease free survival after autologous PBPC transplantation.

For T-cell depletion in HLA-nonidentical bone marrow transplantation of children with malignant diseases, we improved the original lectin/rosetting method described in 1981 by adding anti-CD2/3 coated donor red blood cells to the combination to achieve lectin agglutination in one step. Further improvements in handling led to a shortened and simplified method and better quality of the graft. Five bone marrow grafts prepared with this modified protocol contained a median number of 6 (0-28) x 10(4) T-cells per kg, corresponding to 0.02 (0-0.08)% CD3+ cells and 6 (3.7-10.5) x 10(6) CD34+ cells per kg at a median body-weight of 7 (5-38)kg. The overall recoveries after T-cell depletion were: NC 17 (10-44)%, CD34+ cells 61 (22-100)%, and CFU-GM 55 (29-212)%.

G-CSF mobilized, T-cell-depleted peripheral blood progenitor cells (PBPC) and T-cell-depleted bone marrow (BM) were given to seven children (6 AL, 1 SCID) to prevent severe graft-versus-host-disease (GvHD) as well as graft rejection after transplantation from HLA-nonidentical parental donors. BM was T-cell-depleted by lectin agglutination and E-rosetting. For T-cell-depletion of the PBPC grafts a combination of CD34+ selection with the Ceprate SC immunoadsorption system and a subsequent depletion of CD2+ cells with immunomagnetic Dynabeads was used. The overall recovery was 0.3 (0.1-1.2)% for nucleated cells, 29 (18-45)% for CD3+ cells, respectively. The purity of CD34+ cells was 87 (68-97)% with a 0.3(0.05-0.7)% residual CD3+ T-cell contamination. In spite of the large T-cell number in the PBPC grafts the combination of CD34 positive and subsequent CD2 negative selection achieved a more than 4 log T-cell depletion and prevents severe GvHD even in HLA-nonidentical transplantation. In addition, if a high dose of progenitor cells ensures stable engraftment, this new approach could increase the possibility of wider use of HLA-mismatched family donors for transplantation.

In the past 10 years 201 HLA-identical bone marrow transplantations (BMT) were performed with major ABO incompatibility in 41 (20%) and minor ABO incompatibility in 35 (18%) patients. ABO compatibility between donor and recipient showed no influence on granulocyte and platelet recovery after BMT. Erythrocyte reconstitution was significantly (p < 0.01) delayed for about 1 week in major ABO-incompatible BMT. In addition, a pure red cell aplasia lasting for 2-5 months occurred in 6 out of 21 blood group 0 patients who received transplants of group A. The rate of graft rejection, incidence of graft-versus-host disease as well as the leukemic relapse rate were similar in ABO-compatible and ABO-incompatible BMT. The probability of a 10-year survival after BMT is independent of ABO compatibility between donor and recipient.