"KAIZEN" method realizing implementation of deep-learning models for COVID-19 CT diagnosis in real world hospitals

Numerous COVID-19 diagnostic imaging Artificial Intelligence (AI) studies exist. However, none of their models were of potential clinical use, primarily owing to methodological defects and the lack of implementation considerations for inference. In this study, all development processes of the deep-learning models are performed based on strict criteria of the "KAIZEN checklist", which is proposed based on previous AI development guidelines to overcome the deficiencies mentioned above. We develop and evaluate two binary-classification deep-learning models to triage COVID-19: a slice model examining a Computed Tomography (CT) slice to find COVID-19 lesions; a series model examining a series of CT images to find an infected patient. We collected 2,400,200 CT slices from twelve emergency centers in Japan. Area Under Curve (AUC) and accuracy were calculated for classification performance. The inference time of the system that includes these two models were measured. For validation data, the slice and series models recognized COVID-19 with AUCs and accuracies of 0.989 and 0.982, 95.9% and 93.0% respectively. For test data, the models' AUCs and accuracies were 0.958 and 0.953, 90.0% and 91.4% respectively. The average inference time per case was 2.83 s. Our deep-learning system realizes accuracy and inference speed high enough for practical use. The systems have already been implemented in four hospitals and eight are under progression. We released an application software and implementation code for free in a highly usable state to allow its use in Japan and globally.

A decreasing plasma concentration of a toxicologically active metabolite 9-carboxymethoxymethylguanine after dialysis - A potential new clinical biomarker for improving encephalopathy in patients treated with acyclovir

Although acyclovir is a key drug for the treatment of herpes infections, a consciousness disorder known as "acyclovir encephalopathy" is among its side effects. We encountered a patient with encephalopathy and measured the plasma and cerebrospinal fluid concentrations of acyclovir and its toxicologically active metabolite 9-carboxymethoxymethylguanine (CMMG). Before dialysis, cerebrospinal fluid concentrations of acyclovir and CMMG in this patient with a consciousness disorder were approximately 10% and 1%, respectively, of their plasma concentrations. After 3 days of dialysis, plasma CMMG levels decreased to detectable but below quantitative levels (<0.1 μg/mL), resulting in normal consciousness. These results suggest that decreasing plasma CMMG concentration could be one of clinical biomarkers for improving consciousness in patients with encephalopathy associated with acyclovir.

Pharmacokinetics of loxoprofen in a self-administered overdose in a Japanese patient admitted to hospital

Background

Loxoprofen is a propionic acid derivative and is the most widely prescribed non-steroidal anti-inflammatory drug in Japan. Loxoprofen is generally considered to be relatively nontoxic.

Case presentation

A 33-year-old man (body weight, 55 kg) who intentionally took an overdose of 100 tablets of loxoprofen (6000 mg) as a suicide attempt was emergently admitted to Kyoto Medical Center. On arrival, the patient was suffering disorders of consciousness. His plasma concentrations of loxoprofen and its reduced trans-alcohol metabolite were 52 and 24 μg/mL, 3.7 and 2.3 μg/mL, 0.81 and 0.54 μg/mL, and 0.015 and 0.011 μg/mL, respectively, at 4, 26, 50, and 121 h after the oral overdose. The observed apparent terminal elimination half-life of loxoprofen during days 1 and 2 of hospitalization was in the range 6–12 h, which is several times longer than the reported normal value. This finding implied nonlinearity of loxoprofen pharmacokinetics over the current 100-fold dose range, which could affect the accuracy of values simulated by a simplified physiologically based pharmacokinetic (PBPK) model founded on data from a normal dose of 60 mg. The reasons for the delayed eliminations from plasma of loxoprofen and its trans-alcohol metabolite in this case are uncertain, but slight renal impairment (low eGFR values) developed on the second and third hospital days and could be a causal factor.

Conclusions

Because the patient’s level of consciousness had gradually improved, he was discharged on the fourth day of hospitalization. The virtual plasma exposures of loxoprofen and its reduced trans-alcohol metabolite estimated using the current simplified PBPK model were lower than the measured values in the overdose case. The present results based on drug monitoring data and pharmacokinetic predictions could serve as a useful guide in cases of loxoprofen overdose.