Search

The efficiency of business processes relies on business key performance indicators (Biz-KPIs), that can be negatively impacted by IT failures. Business and IT Observability (BizITObs) data fuses both Biz-KPIs and IT event channels together as multivariate time series data. Forecasting Biz-KPIs in advance can enhance efficiency and revenue through proactive corrective measures. However, BizITObs data generally exhibit both useful and noisy inter-channel interactions between Biz-KPIs and IT events that need to be effectively decoupled. This leads to suboptimal forecasting performance when existing multivariate forecasting models are employed. To address this, we introduce AutoMixer, a time-series Foundation Model (FM) approach, grounded on the novel technique of channel-compressed pretrain and finetune workflows. AutoMixer leverages an AutoEncoder for channel-compressed pretraining and integrates it with the advanced TSMixer model for multivariate time series forecasting. This fusion greatly enhances the potency of TSMixer for accurate forecasts and also generalizes well across several downstream tasks. Through detailed experiments and dashboard analytics, we show AutoMixer's capability to consistently improve the Biz-KPI's forecasting accuracy (by 11-15\%) which directly translates to actionable business insights., Comment: Accepted in the Thirty-Sixth Annual Conference on Innovative Applications of Artificial Intelligence (IAAI-24)

Kidney exchange programs have been developed to overcome the compatibility challenges for the patient with incompatible donors in kidney transplantation. A registry of such incompatible donor-recipient pairs is created, and compatibility is achieved through the exchange of donors. Single-center kidney exchange registries do not have a significant pool size for achieving the full benefits of the kidney exchange program. Thus multi-registry exchange transplants seem to be a natural way forward. There are various challenges in multi-registry exchange programs, such as different constraints for each registry, varying bound on cycle lengths, and data sharing among registries. Researchers have proposed different merging mechanisms with a sequential merger, full merger with fair allocations, and full merger without any restrictions. To form a stable multi-registry sharing mechanism, registries should be individually rational both in the long run as well as in the short run. In this paper, an Integer Programming model for a multi-registry exchange program has been proposed with individual rationality constraints for each registry. A simulation study is conducted to compare the benefits of multi-registry programs with individual registry allocations on Indian data, and multiple analyses have been done. It was concluded that registries with lower arrival rates could benefit 7-9\% in terms of the number of transplants and 4-6\% in terms of quality of matches in a typical kidney exchange registry. Registry with more hard to match patients will benefit significantly larger than registries with more easy to match patients.

Kidney exchange programs have been developed to find compatible kidneys for recipients who have incompatible donors. On the other hand, patients who do not have a living donor depend upon deceased donor (DD) donations to get a kidney transplant. Currently, in India a deceased donor donates kidneys directly to a deceased donor wait-list. The idea of initiating an exchange chain starting from a deceased donor kidney is proposed in a few articles (and recently executed in Italy), but no mathematical formulation has been given for this merger. We have introduced an integer programming formulation that creates deceased donor initiated chains, considering both paired exchange registry and deceased donor allocations simultaneously. There is a possibility of overlap between paired exchange registry and deceased donor wait-list registry data as recipients can register for both registries independently. This has also been addressed in the paper. A long term simulation study is done to analyze the gain of these DD initiated chains over time. It suggests that even with small numbers of deceased donors, these chains can increases the number of potential transplants significantly. Also, waiting time and dropout rates in the merged registry decreases substantially., Comment: 23 pages, manuscript still in progress