The Problem of Fairness in Synthetic Healthcare Data

Access to healthcare data such as electronic health records (EHR) is often restricted by laws established to protect patient privacy. These restrictions hinder the reproducibility of existing results based on private healthcare data and also limit new research. Synthetically-generated healthcare data solve this problem by preserving privacy and enabling researchers and policymakers to drive decisions and methods based on realistic data. Healthcare data can include information about multiple in- and out- patient visits of patients, making it a time-series dataset which is often influenced by protected attributes like age, gender, race etc. The COVID-19 pandemic has exacerbated health inequities, with certain subgroups experiencing poorer outcomes and less access to healthcare. To combat these inequities, synthetic data must “fairly” represent diverse minority subgroups such that the conclusions drawn on synthetic data are correct and the results can be generalized to real data. In this article, we develop two fairness metrics for synthetic data, and analyze all subgroups defined by protected attributes to analyze the bias in three published synthetic research datasets. These covariate-level disparity metrics revealed that synthetic data may not be representative at the univariate and multivariate subgroup-levels and thus, fairness should be addressed when developing data generation methods. We discuss the need for measuring fairness in synthetic healthcare data to enable the development of robust machine learning models to create more equitable synthetic healthcare datasets.     

Xanthates as Chain‐Transfer Agents in Controlled Radical Polymerization (MADIX): Structural Effect of the O‐Alkyl Group

The capability of three chain‐transfer agents, O‐alkyl‐S‐(1‐ethoxycarbonyl)ethyl xanthates (CH₃CHCO₂C₂H₅)S(CS)OZ′, to control the free‐radical polymerization of styrene and ethyl acrylate by the MADIX process was examined. The reactivity of the xanthates varied according to the following trend: Z′  CH₂CH₃ < CH₂CF₃ < CH[P(O)(OEt)₂]CF₃. This change in reactivity allowed a lowering of the polydispersity index from 2.0 for Z′  CH₂CH₃ to 1.15 for Z′  CH[P(O)(OEt)₂]CF₃ in the case of the polymerization of styrene.

Evolution of M _w/M _n with conversion during the polymerization of ethyl acrylate in the presence of xanthates X₁ , X₂ and X₃ . Reaction conditions: [EA]₀ = 4.6 M , [X]₀ = 5.75 × 10⁻² M , [AIBN]₀ = 1.72 × 10⁻³ M ; T = 80 °C ; solvent: toluene.     

Lichenysin

The lipopeptide lichenysin (cyclo-[L-Gln1→D-Leu2→L-Leu3→L-Val4→L-Asp5→D-Leu6→L-Ile7-β-OH fatty acid]) produced by Bacillus licheniformis structurally resembles surfactin from Bacillus subtilis. The main difference is the presence of a glutaminyl residue in position 1 of the peptide sequence in place of glutamic acid in surfactin. This local variation causes significant changes in the properties of the molecule compared to surfactin. Lichenysin has a higher surfactant power, the critical micellar concentration (c.m.c.) being strongly reduced from 220 to 22 μM and a much higher hemolytic activity because 100% hemolysis was observed with only 15 μM instead of 200 μM. Lichenysin is also a better chelating agent because its association constants with Ca²⁺ and Mg²⁺ are increased by a factor of 4 and 16, respectively. This effect is assigned to an increase in the accessibility of the carboxyl group to cations owing to a change in the side chain topology induced by the Glu/Gln exchange. Additionally, the propensity of the lipopeptide for extensive hydrophobic interactions, as illustrated by its low c.m.c., contributes to further stabilization of the cation and an increase in the partitioning of lichenysin into the erythrocyte membrane. Our data support the formation of a lichensyin-Ca²⁺ complex in a molar ratio of 2:1 instead of 1:1 with surfactin, suggesting an intermolecular salt bridge between two lichenysin molecules. Therefore, when Ca²⁺ ions are present in the solution, micellization occurs via a dimer assembly, with a possible long-range effect on the spatial arrangement of the micelles or other supramolecular structures. Finally, among all the surfactin peptidic variants so far known, lichenysin is the one for which the three tested activities are the most substantially improved.     

Influence of Strong Transannular N→P Interaction on Acidity of Cyclenphosphine Sulfide

The acid-base behavior of cyclenphosphine sulfide (cyclenPS) is appreciably different from that of cyclamphosphine sulfide (cyclamPS). The cyclenPS shows five acid functionalities compared to four for cyclamPS. The fifth acidic group in cyclenPS corresponds to the formation of a stable amidure in aqueous solution (pK=12.3). This behavior is due to the strong transannular N→P interaction. The deprotonation sequences were established by ³¹P-NMR and confirmed by modelling of cyclenPS.     

Xanthurenic Acid in the Shell Purple Patterns of Crassostrea gigas: First Evidence of an Ommochrome Metabolite in a Mollusk Shell

Ommochromes are one of the least studied groups of natural pigments, frequently confused with melanin and, so far, exclusively found in invertebrates such as cephalopods and butterflies. In this study focused on the purple color of the shells of a mollusk, Crassostrea gigas, the first evidence of a metabolite of ommochromes, xanthurenic acid (XA), was obtained by liquid chromatography combined with mass spectrometry (UPLC-MS). In addition to XA and various porphyrins previously identified, a second group of high molecular weight acid-soluble pigments (HMASP) has been identified with physicochemical and structural characteristics similar to those of ommochromes. In addition, fragmentation of HMASP by tandem mass spectrometry (MS/MS) has revealed a substructure common to XA and ommochromes of the ommatin type. Furthermore, the presence of melanins was excluded by the absence of characteristic by-products among the oxidation residues of HMASP. Altogether, these results show that the purple color of the shells of Crassostrea gigas is a complex association of porphyrins and ommochromes of potentially ommatin or ommin type.     

Electron localization function comparative study of ground state,triplet state,radical anion,and cation in model carbonyl and imine compounds

The modifications of bonding in carbonyl and imine compounds upon excitation, electron attachment, and ionization were studied within the framework of the electron localization function (ELF). The topological analysis of ELF allows a partition of the molecular space into regions having a clear chemical meaning: the basins. The electronic populations of the basins associated with bonding and nonbonding character, as well as the basin spin densities, were calculated at the MP2 level of the quantum mechanical calculation. Good agreement was found with the classical view in terms of mesomeric structures corresponding to the dominant localization of electrons contained in frontier molecular orbitals (MOs). The variations of the basins population were compared to the predictions of MO theory. ©1999 John Wiley & Sons, Inc. J Comput Chem 20: 897–910, 1999     

Enantiomeric separation of acidic drugs by capillary electrophoresis using a combination of charged and uncharged β-cyclodextrins as chiral selectors

High resolution could be achieved for the enantiomers of acidic drugs, namely, sulindac, fenoprofen, ketoprofen, warfarin, and hexobarbital, in a buffer of pH 3 by the simultaneous addition of uncharged and charged β-cyclodextrin derivatives. The interaction of the analytes with the anionic sulfobutyl ether β-cyclodextrin provides the analytes with an adequate electrophoretic mobility whereas the interaction with various neutral β-cyclodextrins generates high enantioselectivity. Five neutral cyclodextrins, the native β-cyclodextrin, as well as methyl-, dimethyl-, trimethyl- and hydroxypropyl-β-cyclodextrin, were tested to enhance the enantioselectivity of the electrophoretic system. High resolution values and the shortest analysis times for the five drugs tested were achieved in a buffer made of 100 mM phosphoric acid adjusted to pH 3 with triethanolamine and containing dimethyl- or trimethyl-β-cyclodextrin in addition to sulfobutyl ether β-cyclodextrin.