Search

This paper describes a methodology for developing a new confidence metric to improve power grid operator reliance on ML event classifiers. Unlike traditional confidence scores that are generated by the ML, this confidence metric is generated by humans who have spent time studying the performance boundaries of the ML classifier. We refer to this metric as an Expert Derived Confidence (EDC) score. As an initial test of our methodology four participants (3 Subject Matter Experts, 1 Novice) learned the boundaries of an ML’s performance by studying a subset of events in the ML’s training data. Next, the participants rated their confidence in the ML’s ability to classify similar events. The researchers found that all participants’ EDC scores were correlated with the ML’s own uncertainty quantification score and on average EDC scores showed greater confidence in the ML’s ability to correctly classify events when compared to the ML’s own confidence scores. In addition, averaging EDC scores across all participants was the strongest predictor of model performance and predicted performance even after controlling for the ML’s own confidence.

•The responsibility of successful mentorship lays on both participants.•Mentees require training to gain skills to maximize benefits from mentorship.•“Manage up” concepts empower mentees to take an active role in mentorship.

Cartilage injuries are common and may predispose to early accelerated osteoarthritis. Magnetic resonance (MR) imaging is critical in the detection and characterization of acute and chronic cartilage injuries. Several surgical approaches including microfracture, osteochondral allografting, autologous osteochondral transfer system, and autologous chondrocyte implantation have been developed to treat cartilage lesions. The goals of these treatments include relief of symptoms, restoration of joint function, and delay of progression of osteoarthritis. Imaging evaluation by MR imaging plays a crucial role in diagnosis and surgical planning as well as monitoring for postoperative complications.

To address the increasing need for efficient data analysis and decision-making in defense, the U.S. Navy is prioritizing AI/ML systems capable of handling multi-source data and suggesting courses of actions. Historically, many such systems failed due to technical issues and lack of usability or mission relevance. Research on human-AI collaboration aims to create AI systems that better integrate with frontline operators. A recent National Academies of Sciences, Engineering, and Medicine Report (NASEM) outlined 57 research goals, but the U.S. Navy requires a more-focused set of priorities. A workshop with 23 experts from various fields was held at the Naval Information Warfare Center Pacific, resulting in five key research priorities spanning different time frames. This panel discusses the workshop’s findings, highlighting critical questions going into this workshop and those coming out of it. Panel participants come from government, academia, and industry, providing unique perspectives on big questions for human-AI teaming.

Many of the challenges associated with cybersecurity operations are also ripe opportunities for the application of human-machine teaming. Advances in cognitive science, artificial intelligence, and machine learning promise the technology to achieve the level of intelligence and sophistication necessary to create a true intelligent teammate. This panel gathers experts from across the community to discuss the challenges and opportunities for human-machine teaming in cybersecurity operations.

Automation can be unreliable. This makes appropriate trust and reliance difficult to calibrate. One solution to building appropriate trust is to increase automation transparency by displaying information to the operator about the technology’s underlying analytical principles. However, displaying this additional information may increase operator workload. The research and development community must balance the competing demands of providing adequate transparency and keeping operator workload low. To investigate the complex effects of transparency on workload, a modeling approach can be used by computing a measure of processing efficiency called the capacity coefficient. We conducted a study to examine the impact of increasing transparency on operator workload using the capacity coefficient. We present the data from one participant with the goal of demonstrating the utility of the capacity coefficient. We discuss how this participant’s data highlights the inferences possible from capacity analysis for measuring the impact of display design and increased transparency on operator workload.

ROSS is a new legal research tool leveraging artificial intelligence. This tool is somewhat unique both in terms of its interface and underlying functionality. ROSS’s deviation from more traditional systems may have an impact on the adoption of this new technology. The researchers in the current study conducted a mixed methods evaluation of ROSS to assess user experience and likelihood of adoption with a particular emphasis on evaluating user trust in the technology. The researchers conducted simulation interviews with eleven experienced bankruptcy attorneys. In addition, a human-computer trust survey was administered to assess user trust in the technology at various time points throughout the study. This quantitative survey measured both cognitive and affect-based trust and revealed a substantial increase in trust with exposure to the system. Participants were particularly impressed with attributes of the system that are believed to form the cognitive basis of trust, such as the tool’s reliability and understandability. Despite these positive findings, our qualitative assessment of experiential trust was mixed and most users concluded that they would not use ROSS as their primary research tool. This decision may be attributed to a lack of faith in the technology and concerns related to ROSS’s usability. The simulations were conducted between February and April of 2017. It should be noted that ROSS has been updated since this time and certain findings may no longer be representative of the current platform.

Rosie Thomas, Bombay before Bollywood: Film City Fantasies. Albany: State University of New York Press, 2015, 346 pp., US$29.95, ISBN 978-1-4384-5676-8

Rapid allograft infection complicates liver transplantation (LT) in patients with hepatitis C virus (HCV). Pegylated interferon‐α and ribavirin therapy after LT has significant toxicity and limited efficacy. The effect of a human monoclonal antibody targeting the HCV E2 glycoprotein (MBL‐HCV1) on viral clearance was examined in a randomized, double‐blind, placebo‐controlled pilot study in patients infected with HCV genotype 1a undergoing LT. Subjects received 11 infusions of 50 mg/kg MBL‐HCV1 (n = 6) or placebo (n = 5) intravenously with three infusions on day of transplant, a single infusion on days 1 through 7 and one infusion on day 14 after LT. MBL‐HCV1 was well‐tolerated and reduced viral load for a period ranging from 7 to 28 days. Median change in viral load (log10IU/mL) from baseline was significantly greater (p = 0.02) for the antibody‐treated group (range −3.07 to −3.34) compared to placebo group (range −0.331 to −1.01) on days 3 through 6 posttransplant. MBL‐HCV1 treatment significantly delayed median time to viral rebound compared to placebo treatment (18.7 days vs. 2.4 days, p < 0.001). As with other HCV monotherapies, antibody‐treated subjects had resistance‐associated variants at the time of viral rebound. A combination study of MBL‐HCV1 with a direct‐acting antiviral is underway.

Rapid allograft infection complicates liver transplantation (LT) in patients with hepatitis C virus (HCV). Pegylated interferon-α and ribavirin therapy after LT has significant toxicity and limited efficacy. The effect of a human monoclonal antibody targeting the HCV E2 glycoprotein (MBL-HCV1) on viral clearance was examined in a randomized, double-blind, placebo-controlled pilot study in patients infected with HCV genotype 1a undergoing LT. Subjects received 11 infusions of 50 mg/kg MBL-HCV1 (n = 6) or placebo (n = 5) intravenously with three infusions on day of transplant, a single infusion on days 1 through 7 and one infusion on day 14 after LT. MBL-HCV1 was well-tolerated and reduced viral load for a period ranging from 7 to 28 days. Median change in viral load (log10IU/mL) from baseline was significantly greater (p = 0.02) for the antibody-treated group (range −3.07 to −3.34) compared to placebo group (range −0.331 to −1.01) on days 3 through 6 posttransplant. MBL-HCV1 treatment significantly delayed median time to viral rebound compared to placebo treatment (18.7 days vs. 2.4 days, p < 0.001). As with other HCV monotherapies, antibody-treated subjects had resistance-associated variants at the time of viral rebound. A combination study of MBL-HCV1 with a direct-acting antiviral is underway.

The Institute of Medicine has called for physician education as a key step in medical error prevention. In our 2002 national survey, pediatric resident education about medical error prevention was sporadic. We sought to describe the amount and type of pediatric resident training about medical errors and to assess the change in training since 2002.

We review extensive data in patients with coronary heart disease demonstrating the benefits of exercise training (ET) programs with cardiac rehabilitation on various cardiovascular (CV) factors, including exercise capacity, plasma lipids, obesity indices, diabetes mellitus/metabolic syndrome, inflammation, blood rheology, autonomic function, indices of ventricular repolarization dispersion, psychological risk factors, quality of life, and overall major CV morbidity and mortality. In addition, we discuss the positive outcomes of ET programs in patients with peripheral arterial disease, including benefits on objective assessments of claudication and walking distance, as well as CV risk factors and overall morbidity and mortality.

Signal Detection Theory (SDT) has proven to be an effective tool for analyzing perceptual and decision-making performance in many different areas. The main contribution of SDT is its capability to separate the discrimination process from the decision process by distinguishing between measures of sensitivity and measures of response bias. However, the usefulness of the traditional SDT measures of sensitivity, d', and response bias, c, is questionable in applied settings where observers do not make decisions based on an underlying psychophysical continuum. The purpose of our research was to derive alternative measures of sensitivity and response bias, overcoming the limitations of the traditional SDT model. Results from a Monte Carlo simulation showed support for our more parsimonious model.

Alarm system operators often manage multiple alarm systems concurrently. Because such situations often accompany cascading events, it is important to know how operators sequence responses. We examined how the relative reliability and priority of two concurrent alarms affected alarm gauge reset patterns. We hypothesized that operators would respond first to an alarm with higher reliability or higher priority when the other variable was held constant. We also expected participants to respond to alarms with higher priority first when they occurred at the same time as a high-reliability alarm. Sixty-one Old Dominion University undergraduates performed a tracking task and responded to gauge fluctuations (alarms). A between subjects ANOVA revealed that participants responded to alarms with higher priority first when reliability was constant, and to higher reliability alarms first when priority was constant. Our results suggest that relative priority and reliability may be useful parameters to control in complex task situations.

Presents the notes and guitar chords of the song "A Gunshot to the Head of Trepidation," by Trivium.

As the variety and capabilities of cockpit weather displays have increased, weather deviation decisions have become more complex. Important issues include conflicting or outdated weather information, and teamed reactivity of crews to weather displays. We investigated the influence of onboard and NEXRAD agreement, range to the simulated potential weather event, and the pilot flying on collective weather deviation decisions. Twelve pilot-copilot teams flew a simulated route while reacting to weather events presented in two graphical formats on a separate visual display. Results showed that pilots often chose to deviate from weather rather than confront it. When onboard and NEXRAD displays did not agree, flight crews reacted by trusting the onboard system more but using the NEXRAD system as a backup. These results suggest that future weather displays should exploit existing benefits of NEXRAD presentation for situation awareness while retaining the display structure and logic inherent in the onboard system.

The researchers investigated the effects of a likelihood alarm display (LAD) on perceptions of workload and situation awareness (SA) during varying degrees of taskload. Twenty-four psychology students reacted to alarms while performing a complex primary task. The researchers examined participants performances during four experimental sessions. Two conditions of alarm display (binary and likelihood) and task load (low and high) were manipulated within groups. In addition, alarm display order (binary first or likelihood first) and task load order (low first or high first) were examined between groups. Results showed greater SA and reduced workload when participants used a likelihood alarm display and when they experienced low task load. Significant order effects also occurred. This study demonstrated the positive effects of a likelihood alarm display on perceptions of workload and SA, and suggests that LAD use may reduce workload and enhance SA.

Researchers have shown that increasing the perceived urgency of nonverbal auditory alarms can meliorate alarm response performance degradation due to the cry wolf effect. We conducted two experiments to examine the effects of verbal alarm urgency on the cry wolf effect. Experiment 1 revealed that participants responded significantly less often and correctly rejected more urgent alarms. Experiment 2 revealed that the results in Experiment 1 were contingent on the presence of additional task critical information. Response degradation due to the cry wolf effect was not moderated by alarm urgency in either study. However, the results suggest that performance may still benefit from increased alarm urgency provided additional task critical information is available. These results can be utilized to design verbal alarm systems that evoke more appropriate reactions to false alarms.

Researchers have begun examining variables that may moderate the degrading effects of alarm mistrust on alarm reaction performance. We examined alarm systems of varying reliability levels (60 or 80 percent true alarms) that generated either short or long duration alarms. We studied the impact of these variables on participant response frequency and perception of signal validity. The researchers sampled 40 Old Dominion University psychology students. We predicted that participants would rate long duration alarm signals as more representative of a valid signal. We also believed that participants would use the representativeness heuristic as a response strategy. The results partially supported our hypotheses. Participants rated the long duration signals as significantly more representative of a valid signal (p<.001). However, although participants responded significantly more often to long duration alarms (p<.01), participants in the 80% reliability group responded to significantly more alarms than those in the 60% reliability group (p=.02). The performance results are inconsistent with our hypothesis. Although participants perceived the long duration alarms to be more representative than the short, they did not base their response strategy entirely on alarm representativness. Despite these unexpected findings, the results suggest that alarm signal duration has a significant impact on alarm response frequency. Therefore, designers of complex systems may be able to moderate response degradation due to unreliability by designing systems that generate long duration alarm stimuli.

Previous research has indicated that flight crew communication problems cause the majority of human errors in aircraft cockpits. Such problems may further degrade reaction decisions to unreliable alarm signals. In this research we examine how leadership style and workload affect collective reactions to unreliable alarms. Seventeen psychology student dyads from Old Dominion University were randomly assigned to act as pilots or copilots while performing complex psychomotor and alarm response tasks. Dyad leaders (pilots) were assigned to exhibit autocratic or participative leadership behaviors. Dyads reacted to ten alarms in each condition and were randomly assigned to a low or high psychomotor task workload group. Results indicated that dyads in the low workload group reacted to alarms significantly faster than dyads in the high workload group. Also, copilots indicated higher leadership satisfaction in the participative style. These findings suggest that alarm designers consider workload when implementing alarm systems, and that pilots should be trained to exhibit participative leadership qualities.

USDA APHIS and Moscamed have developed an eradication program for the Mediterranean fruit fly, Ceratitis capitata (Wiedemann), in North and Central America. Currently, the primary chemical used to control fruit flies in Central America is spinosad, Success 0.02 CB$^{{\rm TM}}$ (GF-120). Central America, and Guatemala in particular, has a thriving apiculture industry. Many Guatemalan beekeepers believe that Success is toxic to honey bees, Apis mellifera L., and that its use has resulted in a decrease in bee populations and subsequent honey production. Continuous, limited, and topical exposure studies demonstrated that Success is toxic to honey bees at varying levels, depending upon exposure and drying time, as demonstrated by 24 h bioassays. Future studies should evaluate the effects of Success in the field on foraging honey bees and brood health during a season long Medfly control program to determine possible honeybee impacts.

An experimental study was conducted to investigate the potential use of intravascular gene therapy with adenovirusmediated Ad vascular endothelial growth factor VEGF or angiopoietinl Ang1 for the enhancement of muscle flap perfusion and to evaluate the effect of therapy on microcirculatory hemodynamics and microvascular permeability in vivo by using a cremaster muscle flap model in the rat. The cremaster tube flap was left intact after isolation of the pudoepigastric pedicle. A total of 90 male SpragueDawley rats were divided into five groups of 18 each, according to the type of intraarterial treatment. Control flaps received phosphatebuffered saline. Group 2 the control gene encoding green fluorescent protein, AdGFP served as the adenovirus control. In Groups 3, 4, and 5, flaps were pretreated with AdVEGF, AdAng1, and AdAng1 AdVEGF, respectively. Flaps were preserved in a subcutaneous pocket in the hindlimb for evaluation of functional capillary density and microvascular permeability indices at 3, 7, and 14 days by intravital microscopy system. At day 7 and 14, AdVEGF, AdAng1, and combined treatment groups showed significantly higher numbers of capillary densities when compared with control and AdGFP groups p< 0.05. At day 14, AdVEGF was the superior treatment group compared with AdAng1 and AdVEGF AdAng1 p< 0.05. Overall, there was a linear increase in the number of functional capillaries in all treatment groups p< 0.05. At day 3 after AdAngl therapy, a significantly lower permeability index was found when compared with AdVEGF AdAng1 and AdVEGF alone treatment p< 0.05. At day 7, the AdVEGF group had the highest score of permeability index compared with control, combined, and AdAng1 groups p< 0.05. Histologic evaluation of muscle flaps demonstrated mild focal inflammation. There was evidence of mild vasculitis in all flaps except control muscles. Intravascular angiogenic therapy with AdVEGF or AdAng1 was technically feasible, as demonstrated by expression of the control gene, GFP, along the vascular tree. All treatment groups increased perfusion of the muscle flap over a period of 14 days, indicating a longlasting effect of gene therapy. Ang1 alone or in combination with VEGF was as effective as VEGF alone in augmenting muscle perfusion with more stable vessels 1 week after gene therapy.

The capacity of an adenovirus encoding the mature form of vascular endothelial growth factor (VEGF)–D, VEGF-DΔNΔC, to induce angiogenesis, lymphangiogenesis, or both was analyzed in 2 distinct in vivo models. We first demonstrated in vitro that VEGF-DΔNΔC encoded by the adenovirus (Ad-VEGF-DΔNΔC) is capable of inducing endothelial cell proliferation and migration and that the latter response is primarily mediated by VEGF receptor-2 (VEGFR-2). Second, we characterized a new in vivo model for assessing experimental angiogenesis, the rat cremaster muscle, which permits live videomicroscopy and quantitation of functional blood vessels. In this model, a proangiogenic effect of Ad-VEGF-DΔNΔC was evident as early as 5 days after injection. Immunohistochemical analysis of the cremaster muscle demonstrated that neovascularization induced by Ad-VEGF-DΔNΔC and by Ad-VEGF-A165 (an adenovirus encoding the 165 isoform of VEGF-A) was composed primarily of laminin and VEGFR-2–positive vessels containing red blood cells, thus indicating a predominantly angiogenic response. In a skin model, Ad-VEGF-DΔNΔC induced angiogenesis and lymphangiogenesis, as indicated by staining with laminin, VEGFR-2, and VEGFR-3, whereas Ad-VEGF-A165 stimulated the selective growth of blood vessels. These data suggest that the biologic effects of VEGF-D are tissue-specific and dependent on the abundance of blood vessels and lymphatics expressing the receptors for VEGF-D in a given tissue. The capacity of Ad-VEGF-DΔNΔC to induce endothelial cell proliferation, angiogenesis, and lymphangiogenesis demonstrates that its potential usefulness for the treatment of coronary artery disease, cerebral ischemia, peripheral vascular disease, restenosis, and tissue edema should be tested in preclinical models.

The capacity of an adenovirus encoding the mature form of vascular endothelial growth factor (VEGF)–D, VEGF-DΔNΔC, to induce angiogenesis, lymphangiogenesis, or both was analyzed in 2 distinct in vivo models. We first demonstrated in vitro that VEGF-DΔNΔC encoded by the adenovirus (Ad-VEGF-DΔNΔC) is capable of inducing endothelial cell proliferation and migration and that the latter response is primarily mediated by VEGF receptor-2 (VEGFR-2). Second, we characterized a new in vivo model for assessing experimental angiogenesis, the rat cremaster muscle, which permits live videomicroscopy and quantitation of functional blood vessels. In this model, a proangiogenic effect of Ad-VEGF-DΔNΔC was evident as early as 5 days after injection. Immunohistochemical analysis of the cremaster muscle demonstrated that neovascularization induced by Ad-VEGF-DΔNΔC and by Ad-VEGF-A165(an adenovirus encoding the 165 isoform of VEGF-A) was composed primarily of laminin and VEGFR-2–positive vessels containing red blood cells, thus indicating a predominantly angiogenic response. In a skin model, Ad-VEGF-DΔNΔC induced angiogenesis and lymphangiogenesis, as indicated by staining with laminin, VEGFR-2, and VEGFR-3, whereas Ad-VEGF-A165stimulated the selective growth of blood vessels. These data suggest that the biologic effects of VEGF-D are tissue-specific and dependent on the abundance of blood vessels and lymphatics expressing the receptors for VEGF-D in a given tissue. The capacity of Ad-VEGF-DΔNΔC to induce endothelial cell proliferation, angiogenesis, and lymphangiogenesis demonstrates that its potential usefulness for the treatment of coronary artery disease, cerebral ischemia, peripheral vascular disease, restenosis, and tissue edema should be tested in preclinical models.

Inflammatory breast cancer (IBC) is a specific type of breast tumor that generally has a poor prognosis, in spite of recent advances in treatment. In the present study, semiquantitative reverse transcriptase polymerase chain reaction examination of resected specimens showed that angiogenic factors, not lymphangiogenic factors, are overexpressed in IBC tumors, compared with non-IBC tumors. Immunohistochemical analysis of the specimens revealed a significantly higher population of tumor-infiltrating (TI) endothelial cells (ECs) or endothelial precursor cells (EPCs) in tumor-associated stroma of IBC specimens than in non-IBC specimens. In a previous study, we examined the phenotype of host cells in response to transplanted IBC cells, using an established human IBC xenograft model (WIBC-9) (Shirakawa et al., Cancer Res 2001;61:445–51). The data obtained in that study are consistent with the findings of the present study. To explore the therapeutic potential of blocking vascular endothelial growth factor (VEGF) and angiopoietin (Ang) pathways in IBC, established vectors encoding soluble Flt-1 (sFlt-1) and soluble Tie2 (sTie2) were injected directly into WIBC-9. Both vectors produced growth inhibition ratios of WIBC-9 that were significantly higher than those of a non-IBC xenograft (MC-5). Also, both vectors suppressed WIBC-9 lung metastases. The efficacy correlated with the number of TI ECs/EPCs, which was determined by fluorescence-activated cell sorting. These ECs/EPCs incorporated acetylated lipoprotein and were integrated within a HUVEC monolayer in vitro culture on day 5. © 2002 Wiley-Liss, Inc.

A novel approach to treat ischemic tissues by using gene therapy has recently been introduced on the basis of the angiogenic potential of certain growth factors. The authors investigated the effect of adenovirusmediated gene therapy with vascular endothelial growth factor VEGF delivered into the subdermal space to treat compromised skin flaps.For this purpose, the epigastric skin flap model in rats, based solely on the right inferior epigastric vessels, was used. Thirty male SpragueDawley rats were divided into five groups of six rats each. Viral transfection with 108plaqueforming units was performed 2 days before the epigastric flap elevation. Rats received subdermal injections of adenovirus encoding VEGF AdVEGF or green fluorescent protein AdGFP as treatment control. Another set of animals n6 received no injections and were designated as control. To determine whether site of injection had an impact on flap viability, injections were given into the predicted local ischemic area AdVEGF local, n6 AdGFP local, n6 and into the midline of the flap AdVEGF midline, n6 AdGFP midline, n6. A flap measuring 8 × 8 cm was outlined on the abdominal skin extending from the xiphoid process proximally and the pubic region distally, to the anterior axillary lines bilaterally. Then, the epigastric flap was elevated as an island on the right inferior epigastric vessels and sutured back to its bed. Flap viability was evaluated at 7 and 14 days after the first operation. The epigastric flaps were scanned to the computer and areas of hypoxic andor necrotic zones relative to total flap surface area were measured and expressed as percentages by using Image Pro Plus software. Specimens were taken for histologic evaluation at day 14 before the animals were killed. Combined area of necrotic and hypoxic zones as well as necrotic zone were decreased to 9.7 ± 1.4 percent and 1.4 ± 0.9 percent in AdVEGF local, and 11.8 ± 1.9 percent and 3.5 ± 1.64 percent in AdVEGF midline compared with the control and AdGFP treatment groups control, 23 ± 3.6 percent and 20.1 ± 3.3 percent AdGFP local, 24.8 ± 4.8 percent and 16.2 ±5.9 percent and AdGFP midline, 23.4 ± 6.9 percent and 19.5 ± 7.7 percent p<0.05. Histologic evaluation by light microscopy failed to demonstrate any quantitative difference in vascularity of skin flaps between the treatment groups. In this study, the authors demonstrated that adenovirusmediated gene therapy using VEGF enhanced epigastric skin flap survival, as confirmed by the significant reduction in combined area of necrotic and hypoxic zones of the flap. Compared with the control, both local and midline subdermal injections of AdVEGF showed improvement in overall flap survival by 57.9 and 48.7 percent, respectively. The results of this study raise the possibility of using adenovirusmediated therapeutic angiogenesis for safer flap surgery in highrisk patients. Plast. Reconstr. Surg.109 1986,2002.

Gene therapy to correct defective genes requires efficient gene delivery and long-term gene expression. The available vector systems have not allowed the simultaneous achievement of both goals. We have developed a chimeric viral vector system that incorporates favorable aspects of both adenoviral and retroviral vectors. Adenoviral vectors induce target cells to function as transient retroviral producer cells in vivo. The progeny retroviral vector particles are then able to stably transduce neighboring cells. In this system, the nonintegrative adenoviral vector is rendered functionally integrative via the intermediate generation of a retroviral producer cell. The chimeric vectors may allow realization of the requisite goals for specific gene-therapy applications.

This study examined the effects of full-length p16 gene transfer by recombinant adenovirus on cell growth and on sensitivity to CDDP or ACNU chemotherapies. We developed a recombinant adenovirus expressing the full-length human p16 gene (AxCA-hp16) by the COS-TPC method. AxCA-hp16 was infected into the p16-null human glioma cell line, U251MG. AxCA-hp16 infection inhibited proliferation of U251MG cells. A proliferation assay employing MTT showed that AxCA-hp16 infection induced chemoresistance, preventing CDDP-induced cell death (11- to 15-fold) and ACNU-induced cell death (80- to 92-fold). In the absence of AxCA-hp16, cell death was induced with CDDP or ACNU at 3 to 5 days after treatment, as demonstrated by Trypan-blue exclusion. Flow-cytometric analysis showed that CDDP or ACNU arrested cells in the G2 phase on day 1 and that cells re-entered the cycle on day 3. However, the cells infected with AxCA-hp16 after CDDP or ACNU treatment showed G1 arrest on day 5 after re-entering the cycle from G2 arrest on day 3. The cells infected with AxCA-hp16 before CDDP or ACNU treatment showed G1 arrest over the 5 days after the infection. This study demonstrated that G1 arrest induced with p16-gene expression prevents ACNU- or CDDP-induced cell death. The cell death induced by ACNU and CDDP therefore appears to occur in the phase after the G1/S check point. Int. J. Cancer 77:47–54, 1998.© 1998 Wiley-Liss, Inc.