Evaluation of prediction models for the occurrence of malaria in the state of Amapá, Brazil, 1997-2016: an ecological study

Lima, Marcos Venicius Malveira de; Laporta, Gabriel Zorello

doi:10.1590/S1679-49742021000100007

Table of contents previous current next

Table of contents

Original article • Epidemiol. Serv. Saúde 30 (1) • 2021 • https://doi.org/10.1590/S1679-49742021000100007 linkcopy

Evaluation of prediction models for the occurrence of malaria in the state of Amapá, Brazil, 1997-2016: an ecological study^*

Authorship SCIMAGO INSTITUTIONS RANKINGS

Objective To evaluate the predictive power of different malaria case time-series models in the state of Amapá, Brazil, for the period 1997-2016.

Methods This is an ecological time series study with malaria cases recorded in the state of Amapá. Ten deterministic or stochastic statistical models were used for simulation and testing in 3, 6, and 12 month forecast horizons.

Results The initial test showed that the series is stationary. Deterministic models performed better than stochastic models. The ARIMA model showed absolute errors of less than 2% on the logarithmic scale and relative errors 3.4-5.8 times less than the null model. It was possible to predict future malaria cases 6 and 12 months in advance.

Conclusion The ARIMA model is recommended for predicting future scenarios and for earlier planning in state health services in the Amazon Region.

Time Series Studies; Malaria; Decision Support Techniques; Epidemiological Monitoring; Forecasting

Year	January	February	March	April	May	June	July	August	September	October	November	December	Total
1997	1,997	1,520	1,023	1,096	1,030	1,364	2,108	2,708	3,506	3,592	2,814	2,110	24,868
1998	1,633	1,863	1,566	1,524	1,773	1,899	2,005	2,534	3,001	1,795	1,929	1,047	22,569
1999	1,077	1,124	994	661	809	801	1,594	3,472	4,435	4,963	5,570	3,146	28,646
2000	2,473	2,731	1,632	1,269	1,563	1,714	2,377	4,336	3,767	5,944	4,574	2,898	35,278
2001	2,861	2,510	2,019	1,513	1,638	1,760	1,920	2,496	2,466	2,588	1,731	985	24,487
2002	1,422	1,250	849	671	593	649	971	1,953	2,330	2,174	2,144	1,251	16,257
2003	963	854	704	690	755	953	1,622	1,601	2,118	2,456	2,258	1,677	16,651
2004	1,949	1,972	1,453	987	1,014	1,126	1,408	1,954	1,848	2,731	2,387	1,841	20,670
2005	1,872	2,524	2,214	1,331	1,517	1,546	2,252	3,133	3,416	3,615	2,681	1,958	28,059
2006	2,505	1,500	1,231	1,101	1,582	1,637	2,403	3,393	3,431	4,371	3,750	2,386	29,290
2007	2,527	1,743	1,560	1,314	1,254	1,241	1,801	2,147	2,036	3,119	1,914	1,319	21,975
2008	1,217	957	823	770	872	745	960	1,121	1,721	2,099	2,144	1,702	15,131
2009	1,558	1,133	955	860	1,049	1,036	1,096	1,501	1,816	1,783	1,541	1,176	15,504
2010	1,466	1,143	994	730	892	970	1,384	1,390	1,292	1,553	2,072	1,502	15,388
2011	1,119	872	830	723	923	941	1,348	2,254	2,324	2,639	3,115	1,910	18,998
2012	1,685	1,257	1,040	693	863	917	1,272	1,344	1,234	1,619	1,925	1,432	15,281
2013	1,675	1,401	939	842	710	717	918	1,272	1,445	2,121	2,004	1,250	15,294
2014	1,057	806	599	516	487	516	646	1,201	1,844	2,163	2,199	1,521	13,555
2015	1,504	1,097	767	597	558	635	971	1,234	1,705	1,906	1,606	1,078	13,658
2016	1,138	1,040	667	532	580	673	962	1,296	1,585	1,554	1,321	925	12,273

Prediction time - 12-month time horizon				Prediction time - 6-month time horizon				Prediction time - 3-month time horizon
Models	MAPEa	Relative MAPEa scale	Theil’s U	Models	MAPEa	Relative MAPEa scale	Theil’s U	Models	MAPEa	Relative MAPEa scale	Theil’s U
ETS	1.25	3.76	0.40	ETS	0.76	11.28	0.29	NNETAR	2.47	1.44	0.64
ARIMA	1.39	3.38	0.47	ARIMA	1.47	5.83	0.56	ETS	2.87	1.24	0.82
TBATS	1.57	2.99	0.52	TBATS	1.60	5.36	0.58	TBATS	3.28	1.09	0.91
ELM	1.64	2.87	0.50	STLM	1.99	4.31	0.73	Null	3.56	1.00	1.45
STLM	1.63	2.88	0.57	NNETAR	2.16	3.97	0.76	STLM	3.96	0.90	1.17
MLP	1.66	2.83	0.55	BATS	2.24	3.83	0.93	ARIMA	4.36	0.82	1.27
BATS	1.81	2.60	0.68	StructTS	3.57	2.40	1.29	BATS	4.67	0.76	1.34
NNETAR	3.29	1.43	1.01	MLP	5.24	1.64	1.96	MLP	5.74	0.62	2.08
Nulo	4.70	1.00	1.63	ELM	6.97	1.23	2.53	ELM	5.82	0.61	2.20
StructTS	10.15	0.46	3.75	Null	8.57	1.00	2.88	StructTS	10.44	0.34	3.43

location_on

Secretaria de Vigilância em Saúde e Ambiente - Ministério da Saúde do Brasil SRTVN Quadra 701, Via W5 Norte, Lote D, Edifício P0700, CEP: 70719-040, +55 (61) 3315-3464 - Brasília - DF - Brazil
E-mail: ress.svs@gmail.com

rss_feed Stay informed of issues for this journal through your RSS reader

[TFN1001] a) MAPE: mean absolute percentage error.

[TFN2001] Notes:

[TFN3001] ARIMA, ETS, TBATS, BATS and STLM: deterministic models.

[TFN4001] StructTS, NNETAR, ELM and MLP: stochastic models.

[TFN5001] Null: null model.

Evaluation of prediction models for the occurrence of malaria in the state of Amapá, Brazil, 1997-2016: an ecological study*

Evaluation of prediction models for the occurrence of malaria in the state of Amapá, Brazil, 1997-2016: an ecological study^*