Evaluating the Diagnostic Performance of Large Language Models on Complex Multimodal Medical Cases.

Large language models showed interpretative reasoning in solving diagnostically challenging medical cases.

A new machine learning algorithm with high interpretability for improving the safety and efficiency of thrombolysis for stroke patients: A hospital-based pilot study.

Background: Thrombolysis is the first-line treatment for patients with acute ischemic stroke. Previous studies leveraged machine learning to assist neurologists in selecting patients who could benefit the most from thrombolysis. However, when designing the algorithm, most of the previous algorithms traded interpretability for predictive power, making the algorithms hard to be trusted by neurologists and be used in real clinical practice. Methods: Our proposed algorithm is an advanced version of classical k-nearest neighbors classification algorithm (KNN). We achieved high interpretability by changing the isotropy in feature space of classical KNN. We leveraged a cohort of 189 patients to prove that our algorithm maintains the interpretability of previous models while in the meantime improving the predictive power when compared with the existing algorithms. The predictive powers of models were assessed by area under the receiver operating characteristic curve (AUC). Results: In terms of interpretability, only onset time, diabetes, and baseline National Institutes of Health Stroke Scale (NIHSS) were statistically significant and their contributions to the final prediction were forced to be proportional to their feature importance values by the rescaling formula we defined. In terms of predictive power, our advanced KNN (AUC 0.88) outperformed the classical KNN (AUC 0.75, p = 0.0192 ). Conclusions: Our preliminary results show that the advanced KNN achieved high AUC and identified consistent significant clinical features as previous clinical trials/observational studies did. This model shows the potential to assist in thrombolysis patient selection for improving the successful rate of thrombolysis.

Artificial intelligence-assisted multistrategy image enhancement of chest X-rays for COVID-19 classification.

Background: The coronavirus disease 2019 (COVID-19) led to a dramatic increase in the number of cases of patients with pneumonia worldwide. In this study, we aimed to develop an AI-assisted multistrategy image enhancement technique for chest X-ray (CXR) images to improve the accuracy of COVID-19 classification. Methods: Our new classification strategy consisted of 3 parts. First, the improved U-Net model with a variational encoder segmented the lung region in the CXR images processed by histogram equalization. Second, the residual net (ResNet) model with multidilated-rate convolution layers was used to suppress the bone signals in the 217 lung-only CXR images. A total of 80% of the available data were allocated for training and validation. The other 20% of the remaining data were used for testing. The enhanced CXR images containing only soft tissue information were obtained. Third, the neural network model with a residual cascade was used for the super-resolution reconstruction of low-resolution bone-suppressed CXR images. The training and testing data consisted of 1,200 and 100 CXR images, respectively. To evaluate the new strategy, improved visual geometry group (VGG)-16 and ResNet-18 models were used for the COVID-19 classification task of 2,767 CXR images. The accuracy of the multistrategy enhanced CXR images was verified through comparative experiments with various enhancement images. In terms of quantitative verification, 8-fold cross-validation was performed on the bone suppression model. In terms of evaluating the COVID-19 classification, the CXR images obtained by the improved method were used to train 2 classification models. Results: Compared with other methods, the CXR images obtained based on the proposed model had better performance in the metrics of peak signal-to-noise ratio and root mean square error. The super-resolution CXR images of bone suppression obtained based on the neural network model were also anatomically close to the real CXR images. Compared with the initial CXR images, the classification accuracy rates of the internal and external testing data on the VGG-16 model increased by 5.09% and 12.81%, respectively, while the values increased by 3.51% and 18.20%, respectively, for the ResNet-18 model. The numerical results were better than those of the single-enhancement, double-enhancement, and no-enhancement CXR images. Conclusions: The multistrategy enhanced CXR images can help to classify COVID-19 more accurately than the other existing methods.

Synthesis of artemiside and its effects in combination with conventional drugs against severe murine malaria.

This research describes the use of novel antimalarial combinations of the new artemisinin derivative artemiside, a 10-alkylamino artemisinin. It is a stable, highly crystalline compound that is economically prepared from dihydroartemisinin in a one-step process. Artemiside activity was more pronounced than that of any antimalarial drug in use, both in Plasmodium falciparum culture and in vivo in a murine malaria model depicting cerebral malaria (CM). In vitro high-throughput testing of artemiside combinations revealed a large number of conventional antimalarial drugs with which it was additive. Following monotherapy in mice, individual drugs reduced parasitemias to nondetectable levels. However, after a period of latency, parasites again were seen and eventually all mice became terminally ill. Treatment with individual drugs did not prevent CM in mice with recrudescent malaria, except for piperaquine at high concentrations. Even when CM was prevented, the mice developed later of severe anemia. In contrast, most of the mice treated with drug combinations survived. A combination of artemiside and mefloquine or piperaquine may confer an optimal result because of the longer half life of both conventional drugs. The use of artemiside combinations revealed a significant safety margin of the effective artemiside doses. Likewise, a combination of 1.3 mg/kg of body weight artemiside and 10 mg/kg piperaquine administered for 3 days from the seventh day postinfection was completely curative. It appears possible to increase drug concentrations in the combination therapy without reaching toxic levels. Using the drug combinations as little as 1 day before the expected death of control animals, we could prevent further parasite development and death due to CM or anemic malaria. Earlier treatment may prevent cognitive dysfunctions which might occur after recovery from CM.

Using a portable monitoring device for diagnosing obstructive sleep apnea in patients with multiple coexisting medical illnesses.

INTRODUCTION: The existing guidelines recommend type III devices should be used in patients without significant comorbidities. OBJECTIVES: This study explored the reliability of using a type III device in patients with significant medical conditions to diagnose sleep apnea. METHODS: Patients had an overnight sleep study conducted simultaneously with both polysomnography (PSG) and a type III (NOX-T3) monitoring device. All patients had stable multiple coexisting medical illnesses without any changes in medications and conditions within 1 month of sleep study. RESULTS: Between July 2019 and March 2020, there were altogether 74 patients recruited with analyzable data. Five major disease groups were identified in the cohort: psychiatric illnesses, stroke, ischemic heart diseases (IHDs), chronic kidney diseases (CKDs), and others. Psychiatric patients with medications were found to have the lowest apnea hypopnea index (AHI) (23.7 per hour) and arousal index (46.6 per hour). The CKD group had the highest mean arousal index (71.4 per hour) and obstructive apnea count (110.2). NOX-T3 respiratory event index (REI) was significantly lower than the PSG AHI (mean REI 31.4 vs. mean AHI: 42.2). The number of patients with no/mild/moderate/severe obstructive sleep apnea (OSA) diagnosed by NOX-T3 and PSG was 7/17/19/31 and 5/11/20/38, respectively. CONCLUSION: NOX-T3 device can reliably diagnose OSA in patients with different stable coexisting medical conditions. There is a tendency for underestimation of the severity of the OSA with NOX-T3 in patients with coexisting medical conditions especially with sedative medications. A positive NOX-T3 reliably diagnoses OSA whereas a negative NOX-T3 result needs to be interpreted with caution.

Artemisone effective against murine cerebral malaria.

BACKGROUND: Artemisinins are the newest class of drug approved for malaria treatment. Due to their unique mechanism of action, rapid effect on Plasmodium, and high efficacy in vivo, artemisinins have become essential components of malaria treatment. Administration of artemisinin derivatives in combination with other anti-plasmodials has become the first-line treatment for uncomplicated falciparum malaria. However, their efficiency in cases of cerebral malaria (CM) remains to be determined. METHODS: The efficacy of several artemisinin derivatives for treatment of experimental CM was evaluated in ICR or C57BL/6 mice infected by Plasmodium berghei ANKA. Both mouse strains serve as murine models for CM. RESULTS: Artemisone was the most efficient drug tested, and could prevent death even when administered at relatively late stages of cerebral pathogenesis. No parasite resistance to artemisone was detected in recrudescence. Co-administration of artemisone together with chloroquine was more effective than monotherapy with either drug, and led to complete cure. Artemiside was even more effective than artemisone, but this substance has yet to be submitted to preclinical toxicological evaluation. CONCLUSIONS: Altogether, the results support the use of artemisone for combined therapy of CM.

Depression: Problem-solving appraisal and self-rated health among Hong Kong Chinese migrant women.

This cross-sectional survey explored the depression status of new migrant women and its relationship with self-rated health in the Hong Kong Chinese context. A convenience sample of 68 migrant women volunteered to participate in the study. The data were collected by using the Problem Solving Inventory, the Center for Epidemiological Studies-Depression questionnaire, and a self-rated health scale. The respondents were found to have a lesser degree of problem-solving appraisal, compared with other populations, and almost half of the volunteers were found to be depressed. Approximately 50% of the women reported their general health as "excellent", "very good", or "good". The Pearson's correlation showed a positive significant correlation between problem-solving appraisal, depression, and self-rated health. The results of the regression analysis showed that family income, self-rated health, and problem-solving confidence are predictive factors of depression. Community nurses could consider using multidisciplinary interventions that focus on life-skills training in order to promote the psychological and general wellness of migrant women in addition to the use of counseling or medication interventions.

Artemisone and artemiside control acute and reactivated toxoplasmosis in a murine model.

Immunocompromised patients are at risk of developing toxoplasmosis, and although chemotherapy is available, standard treatments are often complicated by severe side effects. Artemisinin is a new highly potent antimalarial drug that has activity against Toxoplasma gondii in vitro. However, artemisinin derivatives have previously been ineffective in vivo using a rat model of toxoplasmosis. In the present study, the efficacy of several new artemisinin derivates was investigated for treatment of mice infected with the parasite Toxoplasma gondii. Artemiside and artemisone displayed better inhibition than either artemisinin or artesunate against the parasite in vitro. Artemiside and artemisone treatment controlled parasite replication in vivo, and mice survived the acute infection. In a murine model of reactivated toxoplasmosis, both drugs increased survival, although artemiside was more effective. These results indicate that these newer derivatives of artemisinin may have potential for treatment of toxoplasmosis.

Validation study of a portable monitoring device for identifying OSA in a symptomatic patient population.

BACKGROUND AND OBJECTIVE: Obstructive sleep apnoea syndrome (OSAS) is a common disorder associated with early atherosclerosis, diabetes mellitus, ischaemic heart disease and cerebrovascular disease. The gold standard for confirming OSAS is based on an attended overnight polysomnography (PSG) in a sleep laboratory; however lack of health-care resources creates long waiting times for patient access to this diagnostic test. This study evaluated the ability of a portable sleep-monitoring device to identify patients in Hong Kong with suspected OSAS. METHODS: Patients with symptoms of OSAS were invited to use the ARES (apnoea risk evaluation system) concurrently with an attended inpatient PSG. Several sets of AHI were generated by the ARES provider based on different oxygen desaturation criteria and surrogate parameters of arousal. The results were compared against PSG to determine the optimal sensitivity and specificity. RESULTS: There were 141 patients who completed the study successfully. Results of AHI from the ARES study were presented in the order of different scoring criteria--4% oxygen desaturation alone, obstructive events with 3% oxygen desaturation and obstructive events with 1% desaturation plus surrogate arousal criteria. The sensitivity was 0.84 (95% confidence interval (CI): 0.77-0.90), 0.89 (95% CI: 0.89-0.94) and 0.97 (95% CI: 0.94-0.99), respectively. The specificity was 1, 1 and 0.63 (95% CI: 0.55-0.71), respectively. The receiver operating curve had an area of 0.96, 0.97 and 0.98, respectively. The kappa coefficient varied from 0.24 to 0.55 for agreement of severity between PSG and ARES. The likelihood ratio positive and the likelihood ratio negative were 2.61, infinity, infinity and 0.16, 0.11, 0.05, respectively, in the order of oxygen desaturation described earlier. CONCLUSIONS: The ARES device has reasonable sensitivity and specificity for diagnosing severe OSAS in symptomatic Chinese patients. There is moderate agreement between ARES and PSG in the diagnosis of severe disease, but less agreement in patients with mild/moderate disease.

In vitro study of the anti-cancer effects of artemisone alone or in combination with other chemotherapeutic agents.

PURPOSE: Artemisinins are now established drugs for treatment of malaria. These agents have been shown to possess impressive anti-cancer properties. We have investigated the role of artemisone (ATM), a novel derivative of artemisinin (ART) in a cancer setting both alone and in combination with established chemotherapeutic agents. METHODS: The anti-proliferative effects of ART and ATM were tested on a panel of human cancer cells in vitro using the methylthiazoletetrazolium assay, and the effect on cell cycling established by flow cytometry. Immunoblot analyses were performed to determine effects at the molecular level. Finally, ART and ATM were combined with the common anti-cancer agents oxaliplatin, gemcitabine and thalidomide. RESULTS: ART and ATM caused dose dependent decreases in cell number. ATM was consistently superior to ART, with IC50 s significantly lower in the former. Neither drug caused significant changes to the cell viability (%viable cells >95%), but arrested cell cycling. Blockade was either exclusively at the level of G1, or at all phases of the cell cycle, and associated with reductions in cyclin D1, CDK4 and pRb. Combination studies showed the anti-proliferative effect of ATM was often enhanced by addition of the other drugs, whilst ART exhibited antagonistic properties. CONCLUSIONS: ART and ATM are active in cancer cell lines, with ATM displaying the greater anti-proliferative effect when used alone. ATM also enhances the effects of the above drugs, with ART being less likely to improve activities. Taken together, ATM should be thought of as the ART-derived compound next in line for further study.

Absorption of the novel artemisinin derivatives artemisone and artemiside: potential application of Pheroid™ technology.

Artemisinins have low aqueous solubility that results in poor and erratic absorption upon oral administration. The poor solubility and erratic absorption usually translate to low bioavailability. Artemisinin-based monotherapy and combination therapies are essential for the management and treatment of uncomplicated as well as cerebral malaria. Artemisone and artemiside are novel artemisinin derivatives that have very good antimalarial activities. Pheroid™ technology is a patented drug delivery system which has the ability to entrap, transport and deliver pharmacologically active compounds. Pharmacokinetic models were constructed for artemisone and artemiside in Pheroid™ vesicle formulations. The compounds were administered at a dose of 50.0mg/kg bodyweight to C57 BL/6 mice via an oral gavage tube and blood samples were collected by means of tail-bleeding. Drug concentrations in the samples were determined using an LC/MS/MS method. There was 4.57 times more artemisone in the blood when the drug was entrapped in Pheroid™ vesicles in comparison to the drug only formulation (p < 0.0001). The absorption of artemiside was not dramatically enhanced by the Pheroid™ delivery system.

Facile oxidation of leucomethylene blue and dihydroflavins by artemisinins: relationship with flavoenzyme function and antimalarial mechanism of action.

The antimalarial drug methylene blue (MB) affects the redox behaviour of parasite flavin-dependent disulfide reductases such as glutathione reductase (GR) that control oxidative stress in the malaria parasite. The reduced flavin adenine dinucleotide cofactor FADH(2) initiates reduction to leucomethylene blue (LMB), which is oxidised by oxygen to generate reactive oxygen species (ROS) and MB. MB then acts as a subversive substrate for NADPH normally required to regenerate FADH(2) for enzyme function. The synergism between MB and the peroxidic antimalarial artemisinin derivative artesunate suggests that artemisinins have a complementary mode of action. We find that artemisinins are transformed by LMB generated from MB and ascorbic acid (AA) or N-benzyldihydronicotinamide (BNAH) in situ in aqueous buffer at physiological pH into single electron transfer (SET) rearrangement products or two-electron reduction products, the latter of which dominates with BNAH. Neither AA nor BNAH alone affects the artemisinins. The AA-MB SET reactions are enhanced under aerobic conditions, and the major products obtained here are structurally closely related to one such product already reported to form in an intracellular medium. A ketyl arising via SET with the artemisinin is invoked to explain their formation. Dihydroflavins generated from riboflavin (RF) and FAD by pretreatment with sodium dithionite are rapidly oxidised by artemisinin to the parent flavins. When catalytic amounts of RF, FAD, and other flavins are reduced in situ by excess BNAH or NAD(P)H in the presence of the artemisinins in the aqueous buffer, they are rapidly oxidised to the parent flavins with concomitant formation of two-electron reduction products from the artemisinins; regeneration of the reduced flavin by excess reductant maintains a catalytic cycle until the artemisinin is consumed. In preliminary experiments, we show that NADPH consumption in yeast GR with redox behaviour similar to that of parasite GR is enhanced by artemisinins, especially under aerobic conditions. Recombinant human GR is not affected. Artemisinins thus may act as antimalarial drugs by perturbing the redox balance within the malaria parasite, both by oxidising FADH(2) in parasite GR or other parasite flavoenzymes, and by initiating autoxidation of the dihydroflavin by oxygen with generation of ROS. Reduction of the artemisinin is proposed to occur via hydride transfer from LMB or the dihydroflavin to O1 of the peroxide. This hitherto unrecorded reactivity profile conforms with known structure-activity relationships of artemisinins, is consistent with their known ability to generate ROS in vivo, and explains the synergism between artemisinins and redox-active antimalarial drugs such as MB and doxorubicin. As the artemisinins appear to be relatively inert towards human GR, a putative model that accounts for the selective potency of artemisinins towards the malaria parasite also becomes apparent. Decisively, ferrous iron or carbon-centered free radicals cannot be involved, and the reactivity described herein reconciles disparate observations that are incompatible with the ferrous iron-carbon radical hypothesis for antimalarial mechanism of action. Finally, the urgent enquiry into the emerging resistance of the malaria parasite to artemisinins may now in one part address the possibilities either of structural changes taking place in parasite flavoenzymes that render the flavin cofactor less accessible to artemisinins or of an enhancement in the ability to use intra-erythrocytic human disulfide reductases required for maintenance of parasite redox balance.

Interaction of artemisinins with oxyhemoglobin Hb-FeII, Hb-FeII, carboxyHb-FeII, heme-FeII, and carboxyheme FeII: significance for mode of action and implications for therapy of cerebral malaria.

In line with the enhancement of antimalarial activities of the current clinical artemisinins against parasites cultured under CO, the artemisinins are unaffected in vitro by carboxyhemoglobin (CO-Hb-Fe(II)) or CO-heme-Fe(II), but are competitively decomposed by Hb-Fe(II) or heme-Fe(II). In the latter case, the heme studies are greatly facilitated by solubilization of the heme in aqueous medium by use of arginine. None of the Hb species has an appreciable effect on artemisone, or on other aminoartemisinins, and antimalarial activities are either less affected or remain essentially unchanged against parasites cultured under standard microaerophilic conditions or under CO. The findings not only indicate that artemisinins do not require Hb-Fe(II) or heme-Fe(II) for promotion of antimalarial activity, but are also important in relation to the therapy of severe/complicated or cerebral malaria.

The Fe2+-mediated decomposition, PfATP6 binding, and antimalarial activities of artemisone and other artemisinins: the unlikelihood of C-centered radicals as bioactive intermediates.

The results of Fe(2+)-induced decomposition of the clinically used artemisinins, artemisone, other aminoartemisinins, 10-deoxoartemisinin, and the 4-fluorophenyl derivative have been compared with their antimalarial activities and their ability to inhibit the parasite SERCA PfATP6. The clinical artemisinins and artemisone decompose under aqueous conditions to give mixtures of C radical marker products, carbonyl compounds, and reduction products. The 4-fluorophenyl derivative and aminoartemisinins tend to be inert to aqueous iron(II) sulfate and anhydrous iron(II) acetate. Anhydrous iron(II) bromide enhances formation of the carbonyl compounds and provides a deoxyglycal from DHA and enamines from the aminoartemisinins. Ascorbic acid (AA) accelerates the aqueous Fe(2+)-mediated decompositions, but does not alter product distribution. 4-Oxo-TEMPO intercepts C radicals from a mixture of an antimalaria-active trioxolane, 10-deoxoartemisinin, and anhydrous iron(II) acetate to give trapped products in 73 % yield from the trioxolane, and 3 % from the artemisinin. Artemisone provides a trapped product in 10 % yield. Thus, in line with its structural rigidity, only the trioxolane provides a C radical eminently suited for intermolecular trapping. In contrast, the structural flexibility of the C radicals from the artemisinins allows facile extrusion of Fe(2+) and collapse to benign isomerization products. The propensity towards the formation of radical marker products and intermolecular radical trapping have no relationship with the in vitro antimalarial activities of the artemisinins and trioxolane. Desferrioxamine (DFO) attenuates inhibition of PfATP6 by, and antagonizes antimalarial activity of, the aqueous Fe(2+)-susceptible artemisinins, but has no overt effect on the aqueous Fe(2+)-inert artemisinins. It is concluded that the C radicals cannot be responsible for antimalarial activity and that the Fe(2+)-susceptible artemisinins may be competitively decomposed in aqueous extra- and intracellular compartments by labile Fe(2+), resulting in some attenuation of their antimalarial activities. Interpretations of the roles of DFO and AA in modulating antimalarial activities of the artemisinins, and a comparison with antimalarial properties of simple hydroperoxides and their behavior towards thapsigargin-sensitive SERCA ATPases are presented. The general basis for the exceptional antimalarial activities of artemisinins in relation to the intrinsic activity of the peroxide within the uniquely stressed environment of the malaria parasite is thereby adumbrated.